effectsofbloodpressureloweringonoutcome …...drugs of different pharmacological classes. there is...

Review

Effects of blood pressure lowering on outcomeincidence in hypertension: 4. Effects of variousclasses of antihypertensive drugs ^ Overview andmeta-analyses

Costas Thomopoulosa, Gianfranco Paratib,c, and Alberto Zanchettid,e

Journal of Hypertension 2015, 33:195–211aDepartment of Cardiology, Helena Venizelou Hospital, Athens, Greece, bDepartmentof Cardiovascular, Neural and Metabolic Sciences, San Luca Hospital, IstitutoAuxologico Italiano IRCCS, cDepartment of Health Sciences, University of MilanBicocca, dScientific Direction, Istituto Auxologico Italiano IRCCS and eCentro Inter-universitario di Fisiologia Clinica e Ipertensione, University of Milan, Milan, Italy

Correspondence to Alberto Zanchetti, Professor, Direzione Scientifica, Istituto Aux-ologico Italiano, Via L. Ariosto, 13, I-20145 Milan, Italy. Tel: +39 2 619112237; fax:+39 2 619112901; e-mail: [email protected]

Received 16 September 2014 Revised 8 October 2014 Accepted 8 October 2014

J Hypertens 33:195–211 Copyright � 2015 Wolters Kluwer Health, Inc. All rightsreserved.

Jo

Background and objectives: In 68 randomized controlledtrials (RCTs), blood pressure (BP) lowering was obtained byusing drugs of different classes. We have investigatedwhether BP lowering by any of the major drug classes iseffective in reducing the cardiovascular outcomes.

Methods: A total of 55 RCTs (195 267 individuals) weresuitable for drug-class meta-analyses. Risk ratios and their95% confidence intervals of seven fatal and nonfataloutcomes were estimated by a random-effects model.

Results: Twelve RCTs (48 898 patients) compared adiuretic with no treatment. SBP/DBP differences of about�12/–5 mmHg were accompanied by significant reductionsof all outcomes, including mortality. The same results wereobtained by limiting analyses to eight RCTs using low-dosediuretics. Separate analyses for thiazides, chlorthalidoneand indapamide (all low dose) showed each subclass wasassociated with significant reduction of some majoroutcome. Five RCTs (18 724 patients; SBP/DBP difference–10.5/–7 mmHg) showed beta-blockers significantlyreduced stroke, heart failure and major cardiovascularevents. In RCTs comparing calcium antagonists,angiotensin-converting enzyme (ACE) inhibitors andangiotensin receptor blockers (ARBs) with placebo smallerSBP/DBP differences were achieved, mostly because in themajority of these later RCTs the antihypertensive drug andplacebo were added on a background treatment withother antihypertensive agents. Nonetheless, significantreductions of stroke, major cardiovascular events,cardiovascular and all-cause death were obtained withcalcium antagonists (10 RCTs, 30 359 patients); stroke,coronary heart disease, heart failure and majorcardiovascular events by ACE inhibitors (12 RCTs, 35 707patients); and stroke, heart failure and majorcardiovascular events by ARBs (13 RCTs, 65 256 patients).

Conclusion: BP lowering by all classes of antihypertensivedrugs is accompanied by significant reductions of strokeand major cardiovascular events. This supports the conceptthat reduction of these events is because of BP loweringper se rather than specific drug properties. However,evidence of risk reduction of other events and particularlymortality was obtained so far with some drug classes only.As a result of marked differences in the trial design, total

Copyright © 2015 Wolters Kluwerurnal of Hypertension

cardiovascular risk, SBP/DBP differences and statisticalpower, comparisons of meta-analyses of different drug-specific placebo-controlled RCTs appear unwarranted.

Keywords: angiotensin-converting enzyme inhibitors,angiotensin receptor blockers, beta-blockers, blood-pressure-lowering treatment, calcium antagonists, diuretics,drug class, hypertension, meta-analysis, randomizedcontrolled trials

Abbreviations: ACE, angiotensin-converting enzyme;ARB, angiotensin receptor blocker; BP, blood pressure;BPLTTC, Blood Pressure Lowering Treatment Trialists’Collaboration; CHD, coronary heart disease; CI, confidenceinterval; n, number; NNT, number needed to treat; RCT,randomized controlled trial; RR, risk ratio

INTRODUCTION

We have previously conducted an overview of therandomized controlled trials (RCTs) of bloodpressure (BP) lowering by drugs and identified

68 RCTs responding to the following prespecified criteria:BP-lowering drugs compared with placebo or no treatment,or more intense compared with less intense treatment withthe intention to investigate the effects of BP differences oncardiovascular outcomes (intentional BP-lowering trials) orBP-lowering drugs compared with placebo, often on abackground of baseline antihypertensive therapy, even ifthe trial design was not that of investigating the effects of BPdifferences provided that a between-group difference of at

Health, Inc. All rights reserved.

DOI:10.1097/HJH.0000000000000447

www.jhypertension.com 195

mailto:[email protected]

Thomopoulos et al.

least 2mmHg in either SBP or DBP occurred (noninten-tional BP-lowering trials); enrolling at least 40% hyper-tensive individuals (SBP >140 or DBP >90mmHg orcurrent antihypertensive drugs); exclusion of trials investi-gating acute myocardial infarction, heart failure, acutestroke and patients on dialysis; protocol including measure-ment of at least one type of cardiovascular events amongstprimary or secondary endpoints; BP measured at baselineand follow-up; follow-up of at least 6 months; a minimumof five events during follow-up; randomized allocation totreatments; and publication within 31 December 2013 [1].Meta-analysis of these BP-lowering RCTs showed all majorcardiovascular outcomes were significantly reduced byBP-lowering treatment, most of them in a proportionalway to the extent of the between-group BP difference [1].

In the above trials, BP lowering was obtained by usingdrugs of different pharmacological classes. There is anobvious interest in investigating whether BP lowering byany of the major classes of antihypertensive drugs is effec-tive in reducing all or part of cardiovascular outcomes. Anumber of meta-analyses have approached this problem inthe past [2–13], but none of them has been comprehensiveof all BP-lowering RCTs conducted from 1966 to end of2013, and simultaneously exclusive of RCTs comparingdifferent active regimens (the latter type of trials aim atavoiding, instead of inducing, a between-group BP differ-ence) as well as of RCTs investigating antihypertensivedrugs in conditions different from hypertension (such asmyocardial infarction and heart failure).

METHODS

Trial eligibilityOf the 68 BP-lowering trials [1], those considered for thepresent meta-analyses of drug-related effects had to satisfyone of these additional criteria: randomization to a drug ofa given class in monotherapy (versus placebo or no orusual therapy) or one group randomized to two drugs ofdifferent classes in combination compared with anothergroup randomized to one of those drugs (RCT includedinto the class of the drug present in the combination andabsent in the monotherapy group). Trials were includedin the primary analyses even if randomization occurredon a background of baseline therapy and if drugs of otherclasses could be subsequently added according toprotocol.

Secondary meta-analyses could additionally includeRCTs in which randomization to the active group allowedthe initial drug to be chosen amongst two different classes(e.g. diuretics or beta-blockers) and RCTs in which theactive group was initially randomized to a drug combi-nation (versus placebo). These RCTs were included in thesecondary meta-analyses of two separate drug classes.

Those BP-lowering RCTs in which the active or moreactive group could receive different drugs to the investi-gator’s discretion could not be included in the current meta-analysis.

OutcomesData on seven predetermined outcomes were extracted:stroke (fatal and nonfatal); coronary heart disease (CHD)

Copyright © 2015 Wolters Kluwe196 www.jhypertension.com

events (coronary death and nonfatal myocardial infarction);hospitalized heart failure; major cardiovascular events,composite of stroke and CHD; major cardiovascular events,composite of stroke, CHD and heart failure; cardiovasculardeath and all-cause death. The definition of outcomesreported in the original study was retained, but wheneverpossible transient ischaemic attacks, angina, revasculari-zation procedures and nonhospitalized nonfatal heart fail-ure were excluded.

Statistical analysesRCTs were divided in drug-class-specific groups accordingto the criteria detailed in the trial eligibility section.Statistical analyses were done with the methods describedin the studies reporting previous meta-analyses [1,14,15].Risk ratios and their 95% confidence intervals (CIs) for eachtrial were calculated by the Mantel–Haenszel methodsweighted by patient number and follow-up duration, andcombined using a random-effects model. This model waschosen for all the meta-analyses because it avoids theassumption that participants in the individual trials aresampled from populations in which the intervention hasthe same quantitative effect [8]. However, the proportion ofinconsistency across the studies not explained by chancewas quantified with the I2 and the x2 Q statistics (P> 0.1).Furthermore, the influence of individual trials on pooledeffect sizes was tested by excluding one trial at a time: if thepoint estimate of the combined effect size with a given trialexcluded lay outside the CI of the overall estimate risk withall available trials, the trial in question was considered tohave an excessive influence.

Calculations of the 5-year incidence of each outcome inthe placebo or control group were made, and the 5-yearabsolute risk reductions by BP-lowering treatment calcu-lated, as well as the numbers of patients needed to treat for5 years in order to prevent an outcome (NNT). Standard-ization of risk ratios to a 10/5 mmHg SBP/DBP differencewas also made as described previously [1].

Comprehensive Meta Analysis version 2 (Biostat, Eagle-wood, New Jersey, USA) was used for all the analyses. Ineach meta-analysis, a P value less than 0.05 was consideredto indicate statistical significance; however, this statisticalthreshold should be interpreted with caution because of themultiple comparisons performed.

RESULTS

Trial and patientsAmongst the 68 BP-lowering RCTs identified in our initialsurvey of BP-lowering trials, 55 (195 267 individuals and773 137 individual years) were found suitable for drug-classspecific meta-analyses because they satisfied all the pre-specified criteria (see trial eligibility) [16–73]. The remain-ing 13 RCTs (mostly of more versus less intense BPlowering) could not be included in any drug-class-specificgroup because more intense treatment could be done bydifferent drugs at the choice of the investigator.

Meta-analyses were done for six different drug classes:diuretics, beta-blockers, calcium antagonists, angiotensin-converting enzyme (ACE) inhibitors, angiotensin receptorblockers (ARBs) and centrally acting drugs. An additional

r Health, Inc. All rights reserved.Volume 33 � Number 2 � February 2015

Outcome reductions by different antihypertensives

meta-analysis was done by considering ACE inhibitors andARBs together (renin–angiotensin system blockers).

DiureticsTable 1 lists the 12 RCTs (48 898 patients) suitable for theprimary analysis of placebo-controlled studies with activetreatment initiated with a diuretic and the nine additionalstudies included in a secondary analysis (66 788 patients).In the primary analysis, a SBP/DBP difference of about�13/–5.5 mmHg between diuretic therapy and placebowas accompanied by a significant reduction in all outcomesconsidered (Fig. 1). The relative risks (RRs) of not onlystroke and heart failure were most markedly reduced [–37%(95% CI –28%, –45%) for stroke, –49% (–34%, –61%) forheart failure], but also CHD [–18% (–6%, –27%)], cardio-vascular death [–18% (–10%, –25%)] and all-cause death

Copyright © 2015 Wolters Kluwer

TABLE 1. BP lowering by diuretics

Trial

Proportion ofhypertensivepatients in

the study (%)Baselinetreatment

Drugs compared

Active Contr

Trials for primary analysisVA-NHLBI [16] 100 No Chlorthalidone Placebo

AUSTRALIAN [17] 100 No Chlorthalidone Placebo

HDFP [18,19] 100 No Chlorthalidone Usual c

MRC-mild [20]a 100 No Bendrofluazide Placebo

Oslo [21] 100 No Hydrochlorothiazide Untreat

EWPHE [22] 100 No Hydrochlorothiazide Placebo

and triamterene

SHEP-pilot [23] 100 No Chlorthalidone Placebo

SHEP [24] 100 No Chlorthalidone Placebo

MRC-old [25]a 100 No Hydrochlorothiazideand amiloride

Placebo

PATS [26] 84 No Indapamide Placebo

HYVET-pilot [27]a 100 No Bendroflumethiazide Untreat

HYVET [28] 100 No Indapamide Placebo

Total primary

Additional trials for secondary analysisVA-1 [29] 100 No Hydrochlorothiazide,

reserpine,hydralazinePlacebo

VA-2 [30] 100 No Hydrochlorothiazide,reserpine, hydralazine

Placebo

Carter [31] 100 No Thiazide andmethyldopaor debrisoquine

Untreat

Barraclough [32] 100 No Bendrofluazide ormethyldopa ordebrisoquine

Placebo

HSCSG [33] 100 No Methyclothiazide,deserpidine

Placebo

USPHS [34] 100 No Chlorthalidone,rauwolfia

Placebo

STOP [35] 100 No Hydrochlorothiazide/amiloride oratenololor metoprololor pindolol

Placebo

PROGRESS [36] b >48 Yes Indapamide andperindopril

Placebo

ADVANCE [37] >75 Yes Indapamide andperindopril

Placebo

Total secondary

BP, blood pressure.aIn MRC-mild, MRC-old and HYVET-pilot trials, the arm randomized to a diuretic has been inclubIn PROGRESS, the subgroup randomized to the combination diuretic and ACE inhibitor has beecompared to the matched group randomized to placebo.

Journal of Hypertension

[–11% (–5%, –17%)] were significantly reduced. Inclusionof additional trials in which diuretics were used in associ-ation with other drugs (secondary analysis) did not sub-stantially change the RR reduction. Heterogeneity was lowin the primary analysis, but markedly increased in thesecondary analysis.

Figure 1 also indicates the absolute incidences of cardio-vascular outcomes in the control group: cardiovasculardeath incidence averaged 4.4% in 5 years. Diuretics pre-vented 15 strokes, 24 major cardiovascular events and eightdeaths every 1000 patients treated for 5 years (with NNT of67, 41 and 118, respectively).

Risk ratios standardized to 10/5mmHg SBP/DBP differ-ences, which are slightly lower than those actually found,were for stroke 0.65 (0.57–0.74); CHD 0.86 (0.77–0.96);heart failure 0.55 (0.44–0.69); composite of stroke and CHD


Patientnumber

Follow-up(years)

BaselineBP

(mmHg)

AchievedSBP

(mmHg)

AchievedDBP

(mmHg)

ol SBP DBP Active Control Active Control

1012 1.5 160 95 NR NR 82.9 88

3427 4 157 100.5 NR NR 88.4 94

are 10940 5 159 101.5 131.5 142.5 86 92

12951 5 161 98 136.5 149.3 87.5 93

ed 785 5.5 156 97 131 148 88 98

840 4.7 182 101 149.5 171.7 86.4 94.7

551 2.8 172 75 140.7 157.6 67.7 71.6

4736 4.5 170 76.5 142.5 155.2 68.3 75

3294 5.8 185 90.5 152 167 78.5 85.5

5665 1.8 154 93 143.4 148.7 86.5 88.9

ed 852 1.1 181 100 151.6 174 83.6 94.5

3845 2.1 173 91 144.7 158 78.2 83.1

48898 4.2 163.7 95.0 139.1 150.9 83.5 88.9

143 1.5 186 121 142.6 182 91.9 118.7

380 3.8 164 104 134.9 189.3 86.4 105

ed 99 4 >160 �110 NR NR 106 115

116 2 NR 110 NR NR 89.8 104.2

452 3 164 100 141 166 88 100

389 7 147 99 131.5 147.4 88.4 98.4

1627 2.1 195 102 166 188.3 87.2 96.7

3544 3.9 149 87 133.1 145.2 79.1 84.1

11140 4.3 145 81 134.7 140.3 74.8 77

66788 4.1 160.5 92.6 138.6 150.0 82.1 87.2

ded together with the entire arm randomized to placebo.n included with the exclusion of the subgroup randomized to ACE-inhibitor only, and


Outcome

StrokeCHDHFStroke + CHDStroke + CHD + HFCV DeathAll-cause Death


Trials(n)

1112

612

81212

19211221142020

DifferenceSBP/DBP(mmHg)

–11.9/–5.5–11.9/–5.5–13.7/–5.2–11.9/–5.5–13.9/–5.7–11.9/–5.5–11.9/–5.5

–11.4/–5.2–11.4/–5.2–11.5/–4.7–11.4/–5.2–12.1/–5.1–11.3/–5.2–11.3/–5.2

Treated

524/21414756/21922

98/72841280/21922729/12662727/21922

1400/21922

971/303001135/30866

337/143502106/308661564/19728

990/290961896/29096

Controls

924/264721116/26976

188/69002040/269761343/177671034/269761845/26976

1539/353601564/35922

479/139653103/359222315/248321397/341482457/34148

Absolute1000 pts/5 yearsrisk reduction

(95% CI)

–15 (–18,–12)–7 (–11, –2)

–17 (–20, –12)–21 (–24, –16)–24 (–29, –19)

–8 (–11, –4)–8 (–13, –4)

–19 (–23, –14)–8 (–11, –5)

–18 (–25, –9)–26 (–32, –20)–30 (–39, –20)–10 (–13, –6)–11 (–15, –7)

NNT5 years

(95% CI)

67 (55, 87)148 (91, 474)

60 (50, 84)47 (41, 61)41 (34, 54)

129 (94, 231)118 (77, 258)

53 (44, 70)128 (95, 197)54 (40, 117)39 (32, 49)33 (26, 50)

104 (78, 167)95 (65, 154)

Events(n/patients) RR

(95% CI)

0.63 (0.55–0.72)0.84 (0.74–0.95)0.51 (0.39–0.66)0.76 (0.69–0.83)0.71 (0.65–0.78)0.82 (0.75–0.90)0.89 (0.83–0.95)

0.63 (0.54–0.72)0.83 (0.77–0.89)0.55 (0.37–0.80)0.74 (0.67–0.82)0.71 (0.62–0.81)0.79 (0.72–0.87)0.87 (0.81–0.92)

Outcome risk(% in 5 years)

4.34.23.87.98.64.47.8

5.34.64.49.8

10.64.68.2

RR(95% CI)

Diuretic better Control better

0.3 0.6 1.0 1.3

P(Heterogen)

0.200.260.380.890.380.670.62

0.0030.48

0.0010.04

<0.0010.340.35

Primary analysis

Secondary analysis

IGURE 1 Relative and absolute risk reduction of various outcomes in the trials of blood pressure lowering by diuretics. Primary and secondary analyses include trials listedTable 1. The column ‘Outcome risk’ reports the percentage incidence of each outcome in the control group calculated on a 5-year span. The column ‘Absolute riskduction’ reports the number (and 95% CI) of events prevented every 1000 patients treated for 5 years with the observed RR. CHD, coronary heart disease; CI,

onfidence interval; CV, cardiovascular; HF, heart failure; n, number; NNT, number (and 95% CI) of patients needed to treat for 5 years to prevent one event; pts, patients;(Heterogen), P for the heterogeneity test x2Q; RR, Mantel–Haenszel risk ratios.

Thomopoulos et al.

FinrecP

0.76 (0.73–0.83); composite of stroke, CHD and heart fail-ure 0.76 (0.71–0.82); cardiovascular death 0.84 (0.77–0.91)and all-cause death 0.90 (0.84–0.96).

As doses of diuretics in antihypertensive therapy havechanged through the years, with lower doses being usedcurrently, separate analyses were done of low and highdose diuretics, trials being classified as low dose if themaximum on-treatment dose of diuretic allowed by proto-col (mostly as second step) was equal or less than 50mghydrochlorothiazide or chlorthalidone and equal or lessthan 5mg bendrofluazide or indapamide. Eight trials wereclassified as low dose [21–28] and four as high dose [16–20].All outcomes considered were significantly reduced by low-

Outcome



Trials(n)

8858688

3414244


–12.2/–5.1–12.2/–5.1–14.3/–5.0–12.2/–5.1–14.4/–5.5–12.2/–5.1–12.2/–5.1

–11.5/–5.7–10.4/–5.7

–10.4/–5.7–12.8/–5.5–10.4/–5.7–10.4/–5.7

Treated

391/9911181/9911

95/5563572/9911540/6644453/9911896/9911

133/11503575/12011

708/12011189/6018

274/12011504/12011

Controls

634/10657308/10657

185/5194942/10657

993/7407637/10657

1137/10657

290/15815808/16319

1098/16319410/10360397/16319708/16319

Absolute Risk Reduction1000 pts/5 years

(95% CI)

–28 (–34, –21)–8 (–12, –3)

–21 (–26, –15)–36 (–44, –28)–51 (–60, –39)–16 (–23, –8)–16 (–26, –5)

–9 (–12, –4)–5 (–13, +4)

–16 (–20, –11)–8 (–13, –2)–4 (–9, +3)

–5 (–11, +3)

NNT5 years

(95% CI)

35 (29, 48)127 (83, 374)

48 (39, 69)39 (32, 49)20 (17, 26)

63 (44, 129)61 (38, 220)

110 (81, 256)184 (75, –224)

62 (50, 93)131 (78, 619)

257 (114, –374)192 (89, –385)


(95% CI)

0.68 (0.60–0.77)0.76 (0.63–0.92)0.51 (0.38–0.67)0.70 (0.63–0.77)0.67 (0.61–0.75)0.81 (0.72–0.91)0.89 (0.82–0.97)

0.51 (0.33–0.79)0.89 (0.73–1.09)

0.76 (0.70–0.84)0.81 (0.68–0.96)0.84 (0.64–1.11)0.88 (0.74–1.06)


9.93.44.7

13.016.08.8

15.2

1.95.0

6.74.02.44.3

RR (95% CI)


0.3 0.6 1.0 1.3

P(Heterogen)

0.560.420.330.890.520.880.80

0.0640.13

0.750.840.160.21

A. Low doses

B. High doses

IGURE 2 Relative and absolute risk reduction of various outcomes in the trials of blood pressure lowering by diuretics at low and high doses. Analyses of trials listed inable 1. (a) Low-dose diuretic [21–28] and (b) high-dose diuretic [16–20]. The column ‘Outcome risk’ reports the percentage incidence of each outcome in the controlroup calculated on a 5-year span. The column ‘Absolute risk reduction’ reports the number (and 95% CI) of events prevented every 1000 patients treated for 5 yearsith the observed RR. CHD, coronary heart disease; CI, confidence interval; CV, cardiovascular; HF, heart failure; n, number; NNT, number (and 95% CI) of patientseeded to treat for 5 years to prevent one event; pts, patients; P (Heterogen), P for the heterogeneity test x2Q; RR, Mantel–Haenszel risk ratios.

FTgwn


dose diuretics, with risk ratios quite similar to those calcu-lated in the overall diuretic meta-analysis (Fig. 2a). Onlystroke and the two composite outcomes were significantlyreduced by high-dose diuretics, whereas reductions in CHDand cardiovascular or all-cause mortality did not reachstatistical significance (Fig. 2b). It should be noted, how-ever, that total cardiovascular risk was much lower in high-dose than in low-dose diuretic trials (cardiovascular death4.8%, rather than 17.6% in 10 years) and hence the statisticalpower of the high-dose diuretic meta-analysis was lower.

Separate analyses were also done according to the typeof diuretic used as active drug, limiting the analyses to trialsusing low doses (Fig. 3). Significant reductions of stroke,


Outcome




Trials(n)

4424344

2222222

2212122


–17.3/–8.1–17.3/–8.1–19.3/–9.0–17.3/–8.1–16.4/–7.7–17.3/–8.1–17.3/-8.1

–13.4/–3.9–13.4/–3.9–13.4/–3.9–13.4/–3.9–13.4/–3.9–13.4/–3.9–13.4/–3.9

–8.5/–4.0–8.5/–4.0

–8.5/–4.0

–8.5/–4.0–8.5/–4.0

Treated

74/232989/232919/822

163/2329183/1903163/2329309/2329

107/280858/280854/2808

165/2808219/2808104/2808245/2808

210/477434/4774

244/4774

186/4774342/4774

Controls

193/3442199/3442

24/803392/3442405/3016298/3442495/3442

155/247976/2479

104/2479231/2479335/2479117/2479249/2479

286/473633/4736

319/4736

222/4736393/4736


(95% CI)

–22 (–32, –8)–11 (–24, +8)–6 (–16, +13)–34 (–46, –21)–31 (–46, –15)–17 (–28, –4)–10 (–25, +7)

–25 (–34, –14)–10 (–17, –1)

–23 (–29, –15)–35 (–46, –21)–59 (–71, –43)–10 (–20, +2)–12 (–27, +7)

–34 (–44, –19)0 (–5, +9)

–35 (–46, –18)

–18 (–29, +1)–27 (–44, –2)

NNT5 years

(95% CI)

46 (32, 128)94 (42, –128)179 (62, –74)

29 (22, 49)32 (22, 65)

59 (35, 249)95 (40, –153)

40 (30, 73)97 (60, 980)43 (35, 66)28 (22, 47)17 (14, 23)

96 (50, –476)82 (36, –149)

29 (23, 54)

29 (22, 56)

55 (34, –804)37 (23, 462)


(95% CI)

0.59 (0.41–0.85)0.80 (0.56–1.15)0.81 (0.45–1.46)0.67 (0.56–0.80)0.74 (0.62–0.87)0.79 (0.65–0.95)0.92 (0.81–1.05)

0.63 (0.50–0.80)0.69 (0.49–0.97)0.48 (0.35–0.67)0.65 (0.54–0.79)0.60 (0.51–0.71)0.80 (0.61–1.04)0.89 (0.75–1.06)

0.73 (0.61–0.87)1.02 (0.63–1.65)

0.76 (0.65–0.89)

0.83 (0.69–1.01)0.86 (0.75–0.99)


5.25.32.9

10.211.88.0

13.1

6.93.44.7

10.415.05.2

11.2

15.91.8

17.7

12.321.8

RR (95% CI)


0.3 0.6 1.0 1.5

P(Heterogen)

0.260.230.550.900.650.440.48

0.470.650.600.770.960.750.57

0.980.38

0.80

0.770.46

A. Thiazides

B. Chlorthalidone

C. Indapamide

IGURE 3 Relative and absolute risk reduction of various outcomes in the trials of blood pressure lowering by different classes of diuretics at low dose. Analyses of trialsted in Table 1. (a) Thiazides [21,22,25,27], (b) chlorthalidone [23,24] and (c) indapamide [26,28]. The column ‘Outcome risk’ reports the percentage incidence of eachutcome in the control group calculated on a 5-year span. The column ‘Absolute risk reduction’ reports the number (and 95% CI) of events prevented every 1000 patientseated for 5 years with the observed RR. CHD, coronary heart disease; CI, confidence interval; CV, cardiovascular; HF, heart failure; n, number; NNT, number (and 95%I) of patients needed to treat for 5 years to prevent one event; pts, patients; P (Heterogen), P for the heterogeneity test x2Q; RR, Mantel–Haenszel risk ratios.


FlisotrC

composites of stroke and CHD and stroke, CHD and heartfailure, and cardiovascular death were found with low-dosethiazides [21,22,25,27]; significant reductions of stroke,CHD, heart failure and their composites (but not of car-diovascular or all-cause mortality) with low-dose chlortha-lidone [23,24]; and significant reductions of stroke,composite of stroke and CHD, and all-cause death bylow-dose indapamide [26,28].


TABLE 2. BP lowering by beta-blockers

Trial



Drugs compared

PnActive Control

Trials for primary analysisMRC-mild [20]a 100 No Propranolol Placebo

HEP [38] 100 No Atenolol Untreated

MRC-old [25]a 100 No Atenolol Placebo

TEST [39] 100 No Atenolol Placebo

UKPDS [40]b 100 Yes, low Atenolol Less active

Total primary

Additional trials for secondary analysisSTOP [35]c 100 No Atenolol or

metoprololor pindolol orhydrochloro-thiazide

Placebo

Total secondary

BP, blood pressure.aIn MRC-mild and MRC-old trials, the arm randomized to a beta-blocker has been included togbIn UKPDS, the subgroup randomized to atenolol in the more intense treatment arm has beencIn STOP, 68% of patients in the active arm received a beta-blocker.


Beta-blockersTable 2 lists the five RCTs (18 724 patients) suitable for theprimary analysis of placebo-controlled studies in whichtreatment was initiated with a beta-blocker, and oneadditional trial included in a secondary meta-analysis[Swedish Trial in Old Patients with Hypertension (STOP-Hypertension) trial [35]], in which two-thirds of the patientsreceived treatment with a beta-blocker, but one-third was


atientumber

Follow-up(years)

BaselineBP

(mmHg)

AchievedSBP

(mmHg)

AchievedDBP

(mmHg)

SBP DBP Active Control Active Control

13057 5 161 98 139.7 149.5 86.5 93

884 4.4 196 98 162.1 180.1 77 88

3315 5.8 185 91 154.5 167 77 86.5

720 2.3 163 90 157 161 85 89

748 8.4 160 94 143 154 81 87

18724 5.1 167.1 96.4 144.2 154.7 84.1 91.0

1627 2.1 195 102 166 188.3 87.2 96.7

20351 4.9 169.3 96.8 145.9 157.35 84.3 91.5

ether with the entire arm randomized to placebo.included together with the entire group randomized to less intense treatment.


Outcome



Trials(n)

5525455

6636566


–10.5/–7.0–10.5/–7.0–14.8/–8.7–10.5/–7.0–10.7/–7.1–10.5/–7.0–10.5/–7.0

–11.4/–7.2–11.4/–7.2–18.5/–9.1–11.4/–7.2–11.7/–7.3–11.4/–7.2–11.4/–7.2

Treated

216/6654293/6654

31/777509/6654446/6282260/6654457/6654

245/7466318/7466

50/1589563/7466519/7094277/7466493/7466

Controls

394/12070538/12070

60/855932/12070901/11722465/12070780/12070

447/12885566/12885

99/16701013/128851021/12537

506/12885843/12885


(95% CI)

–7 (–13, –1)–5 (–10, 0)

–28 (–44, –6)–12 (–20, –4)–16 (–27, –2)–6 (–12, +2)–4 (–10, +4)

–10 (–15, –3)–5 (–10, 0)

–34 (–43, –18)–15 (–22, –6)–20 (–31, –7)–9 (–16, 0)

–9 (–17, +1)

NNT5 years

(95% CI)

136 (80, 1041)194 (101, ∞)35 (23, 170)82 (51, 264)64 (37, 451)79 (86, –540)

273 (102, –274)

105 (68, 314)193 (101, ∞)30 (23, 54)

67 (45, 159)49 (32, 136)111 (64, ∞)

118 (59, –764)


(95% CI)

0.77 (0.61–0.97)0.88 (0.77–1.01)0.57 (0.35–0.91)0.84 (0.74–0.95)0.79 (0.64–0.97)0.85 (0.69–1.05)0.94 (0.84–1.06)

0.73 (0.58–0.91)0.88 (0.77–1.01)0.54 (0.39–0.76)0.81 (0.72–0.92)0.75 (0.62–0.91)0.77 (0.60–0.99)0.87 (0.74–1.02)


3.24.36.57.67.43.76.1

3.54.38.07.98.13.96.5

RR (95% CI)

P(Heterogen)

0.120.470.280.20

0.0100.110.34

0.0790.610.490.17

0.0050.0140.070

Primary analysis

Secondary analysis

Beta-blocker better Control better

0.3 0.6 1.0 1.3

FIGURE 4 Relative and absolute risk reduction of various outcomes in the trials of blood pressure lowering by beta-blockers. Primary and secondary analyses include trialslisted in Table 2. The column ‘Outcome risk’ reports the percentage incidence of each outcome in the control group calculated on a 5-year span. The column ‘Absoluterisk reduction’ reports the number (and 95% CI) of events prevented every 1000 patients treated for 5 years with the observed RR. CHD, coronary heart disease; CI,confidence interval; CV, cardiovascular; HF, heart failure; n, number; NNT, number (and 95% CI) of patients needed to treat for 5 years to prevent one event; pts, patients;P (Heterogen), P for the heterogeneity test x2Q; RR, Mantel–Haenszel risk ratios.

Thomopoulos et al.

treated by a diuretic only. In the primary analysis, a SBP/DBP difference of –10.5/–7.0 mmHg between beta-blockertherapy and placebo (or no treatment; Fig. 4) was accom-panied by a significant reduction in stroke [–23% (–3%to –39%)] and in major cardiovascular events [–16% (–5 to–26%) composite of stroke and CHD; –21% (–3 to –36%)composite of stroke, CHD and heart failure]. Only two trialsprovided separate incidences of heart failure, but also therisk of heart failure was found significantly and markedlyreduced [–43% (–9 to –69%)]. Risk ratios were also lowerthan 1.0 for CHD, cardiovascular and all-cause mortality,but the reductions did not achieve statistical significance.Inclusion of the STOP trial [35] in the secondary meta-analysis made RR reduction of cardiovascular death toachieve statistical significance [–23% (–1 to –40%)]. Hetero-geneity reached statistical significance (P< 0.1) only for thecomposite of stroke, CHD and heart failure in the primaryanalysis, but was slightly increased in the secondaryanalysis (Fig. 4).

Absolute incidences of cardiovascular outcomes in thecontrol group are indicated in Fig. 4: cardiovascular deathincidence averaged 3.7% in 5 years. Absolute risk reductioninduced by beta-blocker treatment (primary analysis)amounted to seven strokes and 16 major cardiovascularevents every 1000 patients treated for 5 years (NNT 136 and64, respectively).

Risk ratio standardization to 10/5 mmHg SBP/DBP differ-ences (values close to those actually occurring) reducedonly slightly the original values: stroke 0.80 (0.66–0.97);heart failure 0.73 (0.55–0.95); composite of stroke, CHDand heart failure 0.82 (0.69–0.98); and cardiovascular death0.87 (0.73–1.04).

Calcium antagonistsTable 3 lists the 10 RCTs (30 359 individuals) included in theprimary analysis of placebo-controlled studies in whichactive treatment was based on a calcium-antagonist


(always a dihydropyridine) and two additional trialsincluded in the secondary analysis. In the primary analysis,a –6/–3.4 mmHg SBP/DBP difference between calciumantagonist therapy and placebo was accompanied by asignificant RR reduction of all outcomes except CHD andheart failure: stroke was reduced by –34% (–25 to –42%),major cardiovascular events (composite of stroke and CHD)by –24% (–8 to –38%), cardiovascular death by –18% (–3to –30%) and all-cause death by –13% (–2 to –23%). A 17%reduction in CHD and a 19% reduction in heart failure didnot achieve statistical significance because of the large CIs.Inclusion of two small trials in the secondary analysis couldnot change the risk ratios and their significance (Fig. 5).

Incidence of cardiovascular death in the placebo groupaveraged 4.2% in 5 years. Absolute risk reduction inducedby calcium antagonist therapy (primary analysis) amountedto 18 strokes, 24 major cardiovascular events (composite ofstroke and CHD) and 9 all-cause deaths every 1000 patientstreated for 5 years (NNT 55, 41 and 107, respectively).

Standardization of risk ratios to a 10/5 mmHg SBP/DBPdifference was also calculated, but with a considerableapproximation because of the fact that the real BP reductionwas only about one-half of that used for standardization:standardized risk ratio for stroke was 0.48 (0.38–0.60), forcomposite of stroke and CHD 0.63 (0.45–0.87), for cardio-vascular mortality 0.72 (0.55–0.95) and for all-causemortality 0.80 (0.66–0.97).

At variance with trials using diuretics or beta-blockers, anumber of calcium antagonist trials also included a pro-portion of normotensive individuals and had baseline anti-hypertensive drugs continued during follow-up, with thecalcium antagonist not being used as the initial drug. Asensitivity analysis including the four trials only enrollinghypertensive patients with no or minimal backgroundtherapy [42–44,51] showed significant and marked riskreductions of all outcomes, including those (CHD and heartfailure), the reduction of which did not reach statistical


TABLE 3. BP lowering by calcium antagonists

Trial



Drugs compared

Patientnumber

Follow-up(years)

BaselineBP

(mmHg)

AchievedSBP

(mmHg)

AchievedDBP

(mmHg)

Active Control SBP DBP Active Control Active Control

Trials for primary analysisHunan [42] 100 No Nitrendipine Untreated 2080 4.7 161 99 140.7 148.9 85.2 90.6

Syst-Eur [43] 100 No Nitrendipine Placebo 4695 2 174 86 151.7 160.5 78.7 84.3

Syst-China [44] 100 No Nitrendipine Placebo 2394 3 171 86 150.6 158.6 81.1 84.1

IDNT [45]a 100 Yes Amlodipine Placebo 1136 2.6 158 87 141 144 77 80

Fogari [46]b 100 No Amlodipine andfosinopril

Fosinopril 206 4 NR NR 132.4 142.3 82.3 87.3

NICOLE [47] 40 No Nisoldipine Placebo 819 3 NR NR NR NR NR NR

CAMELOT [48]a 60 Yes Amlodipine Placebo 1318 2 129 78 124.7 129.3 75.2 78.2

ACTION [49] 52 Yes Nifedipine GITS Placebo 7665 4.9 NR NR 130.3 135.7 76.2 79.2

REIN-2 [50] 60 Yes Felodipine andramipril

Ramipril 335 1.6 131 84 130 134 80 82

FEVER [51] 100 Yes, low Felodipine Placebo 9711 3.3 154 91 137.5 142.2 82.4 85

Total primary 30359 159 89 138.6 144.6 79.7 83.2

Additional trials for secondary analysisBENEDICT-A [52]c 57 Yes Verapamil and

trandolaprilPlacebo 600 3.6 151 88 139 142 80 83

DEMAND [53]d 44.2 Yes Delapril andmanidipine

Placebo 253 3.8 148 87 137.2 139.5 80.5 82.9

Total secondary 31212 159 89 138.6 144.5 79.7 83.2

BP, blood pressure.aIn IDNT and CAMELOT, comparison between the arm randomized to amlodipine with the arm randomized to placebo.bIn Fogari, comparison of the arm randomized to the combination amlodipine and fosinopril with the arm randomized to fosinopril only.cIn BENEDICT-A, comparison of the arm randomized to the combination verapamil and trandolapril with placebo. The comparisons verapamil with placebo, and verapamil andtrandolapril versus trandolapril only were not included because the SBP/DBP differences were too small.dIn DEMAND, comparison of the arm randomized to the combination delapril and manidipine with the arm randomized to placebo. The comparison delapril and manidipine withdelapril only was not included because the SBP/DBP difference was too small.


significance in the primary analysis (Fig. 6). This sensitivityanalysis markedly reduced the high heterogeneity betweenthe trials found in the primary analysis (compare Figs. 5and 6).

Angiotensin-converting enzyme inhibitorsTable 4 lists the 12 RCTs and 13 randomized comparisons(35707 patients) in which active treatment was based on an


Outcome



Trials(n)

986868

10

986

106

1011


–6.1/–3.4–5.9/–3.3–5.8/–3.3–5.9/–3.3–5.8/–3.3–5.8/–3.3–6.0/–3.4

–6.1/–3.4–5.9/–3.3–5.8/–3.3–5.8/–3.3–5.8/–3.3–5.7/–3.3–6.0/–3.4

Treated

414/15099491/14059272/13547868/14059

1140/13547360/13819760/15267

414/15099491/14059272/13547881/14485

1140/13547360/14245762/15393

Controls

622/14925528/13885319/13372

1061/138851361/13372

422/13641842/15092

622/14925528/13885319/13372

1086/143121361/13372

427/14068845/15219

Events(n/patients)

(95

0.66 (0.83 (0.81 (0.76 (0.81 (0.82 (0.87 (

0.66 (0.83 (0.81 (0.74 (0.81 (0.81 (0.87 (


5.85.13.210.614.14.27.3

5.85.13.2

10.514.14.17.3

Primary an

Secondary a

FIGURE 5 Relative and absolute risk reduction of various outcomes in the trials of bloodtrials listed in Table 3. The column ‘Outcome risk’ reports the percentage incidence of ea‘Absolute risk reduction’ reports the number (and 95% CI) of events prevented every 10CI, confidence interval; CV, cardiovascular; HF, heart failure; n, number; NNT, number (apatients; P (Heterogen), P for the heterogeneity test x2Q; RR, Mantel–Haenszel risk ratio


ACE inhibitor and which were included in the primaryanalysis. For the secondary analysis, two additional trialsand an additional comparison in two of the trials included inthe primary analysis were considered. All the additionalgroups compared a treatment initiated with an associationof anACE inhibitor and anotherdrugwithplacebo treatment.

In the primary analysis, a SBP/DBP difference of about�4/–2mmHg between ACE-inhibitor therapy and placebo



(95% CI)

–18 (–22, –14)–8 (–17, +3)–6 (–12, +3)–24 (–37, –8)

–26 (–39, –10)–7 (–12, –1)–9 (–16, –1)

–18 (–22, –14)–8 (–17, +3)–6 (–12, +3)

–26 (–38, –11)–26 (–39, –10)

–8 (–12, –1)–9 (–15, –1)

NNT5 years

(95% CI)

55 (45, 73)118 (60, –386)169 (84, –305)

41 (27, 120)39 (26, 102)

137 (85, 802)107 (62, 683)

55 (45, 73)118 (60, –386)169 (84, –305)

39 (27, 88)39 (26, 102)

133 (81, 817)108 (65, 687)

RR% CI)

0.58–0.75)0.65–1.05)0.60–1.10)0.62–0.92)0.71–0.93)0.70–0.97)0.77–0.98)

0.58–0.75)0.65–1.05)0.60–1.10)0.61–0.89)0.71–0.93)0.68–0.97)0.78–0.98)

RR (95% CI)

P(Heterogen)

0.630.0480.0330.0030.0250.290.25

0.630.0480.0330.0030.0250.290.32

alysis

nalysis

Calcium antagonist better Control better

0.3 0.6 1.0 1.3

pressure lowering by calcium antagonists. Primary and secondary analyses includech outcome in the control group calculated on a 5-year span. The column00 patients treated for 5 years with the observed RR. CHD, coronary heart disease;nd 95% CI) of patients needed to treat for 5 years to prevent one event; pts,s.


Outcome


Trials(n)

4333334


–6.5/–3.7–6.3/–3.5–6.3/–3.5–6.3/–3.5–6.3/–3.5–6.3/–3.5–6.5/–3.7

Treated

306/9532113/8492

59/8492382/8492463/8492162/8492344/9532

Controls

466/9348151/8308

84/8308528/8308612/8308221/8308432/9348


(95% CI)

–26 (–31, –17)–7 (–11, –2)–5 (–7, –1)

–27 (–34, –18)–29 (–37, –20)–12 (–16, –6)–16 (–23, –8)

NNT5 years

(95% CI)

39 (32, 58)143 (94, 579)

194 (135, 1091)37 (29, 55)34 (27, 51)

84 (62, 176)61 (44, 123)


(95% CI)

0.63 (0.53–0.76)0.74 (0.58–0.94)0.68 (0.49–0.95)0.71 (0.62–0.81)0.74 (0.66–0.83)0.71 (0.58–0.87)0.77 (0.67–0.89)


7.82.91.910.312.34.67.7

RR (95% CI)

P(Heterogen)

0.250.630.770.360.640.870.68

Sensitivity analysis

Calcium antagonist better Control better

0.3 0.6 1.0 1.3

FIGURE 6 Relative and absolute risk reduction of various outcomes in the trials of blood pressure lowering by calcium antagonists (only hypertensive patients with no orminimal baseline therapy). Sensitivity analysis including some of the trials [42–44,51] of Table 3. The column ‘Outcome risk’ reports the percentage incidence of eachoutcome in the control group calculated on a 5-year span. The column ‘Absolute risk reduction’ reports the number (and 95% CI) of events prevented every 1000 patientstreated for 5 years with the observed RR. CHD, coronary heart disease; CI, confidence interval; CV, cardiovascular; HF, heart failure; n, number; NNT, number (and 95%CI) of patients needed to treat for 5 years to prevent one event; pts, patients; P (Heterogen), P for the heterogeneity test x2Q; RR, Mantel–Haenszel risk ratios.

Thomopoulos et al.

was associated with significant reductions in the RR of alloutcomes, except cardiovascular and all-cause mortalities.Stroke was reduced by –20% (–7 to –31%), heart failure by–21% (–7 to –34%), CHD by –13% (–3 to –21%) and majorcardiovascular events (composite of stroke, CHD and heart

Copyright © 2015 Wolters Kluwe

TABLE 4. BP lowering by angiotensin-converting enzyme inhibitors

Trial



Drugs compared

Active Control

Trials for primary analysisLewis [54] 75.5 Yes Captopril Placebo

AIPRI [55] 82 Yes Benazepril Placebo

UKPDS [40,41]a 100 Yes, low Captopril Placebo

HOPE [42] 46.9 Yes Ramipril Placebo

PROGRESS [36]b >48 Yes Perindopril Placebo

Fogari [46]c 100 No Fosinopril andamlodipine

Amlodipine

HYVET-pilot [27]d 100 No Lisinopril Untreated

CAMELOT [48]e 60 Yes Enalapril Placebo

BENEDICT-A [52]f 57 Yes Trandolapril Placebo

57 Yes Trandolapril andVerapamil

Verapamil

DIABHYCAR [57] 56 Yes Ramipril Placebo

PEACE [58] 45.5 Yes Trandolapril Placebo

DREAM [59] 43.5 Yes Ramipril Placebo

Total primary

Additional trials for secondary analysisPROGRESS [36]b >48 Yes Perindopril and

indapamidePlacebo

BENEDICT-A [52]f 57 Yes Trandolapril andverapamil

Placebo

ADVANCE [37] >75 Yes Perindopril andIndapamide

Placebo

DEMAND [53]g 44.2 Yes Delapril andmanidipine

Placebo

Total secondary

BP, blood pressure.aIn UKPDS, the subgroup randomized to captopril in the more intense treatment arm has beenbIn PROGRESS, the comparison between the subgroup randomized to perindopril monotherapyanalysis, whereas the comparison between the subgroup randomized to the combination perindsecondary analysis.cIn Fogari, comparison of the arm randomized to the combination fosinopril and amlodipine widIn the HYVET pilot trial, the arm randomized to lisinopril has been compared to the entire armeIn CAMELOT, the arm randomized to enalapril has been compared to the arm randomized tofIn BENEDICT-A, the comparisons trandolapril versus placebo, and trandolapril and verapamil vetrandolapril and verapamil with placebo has been added to the secondary analysis (with placebogIn DEMAND, the comparison between the combination delapril and manidipine with placebo hthe SBP/DBP difference was too small.

202 www.jhypertension.com

failure) by –17% (–8 to 25%). A 11% reduction in cardio-vascular mortality and an 8% reduction in all-causemortality did not achieve statistical significance (Fig. 7).Addition of 15 237 patients in the secondary analysis, mostlypatients on a combination of an ACE inhibitor with a

r Health, Inc. All rights reserved.

Patientnumber

Follow-up(years)

BaselineBP

(mmHg)

AchievedSBP

(mmHg)

AchievedDBP

(mmHg)


409 3 138 86 131 133 79.5 82

583 3 143 88 135.3 145.4 82.7 88.9

790 8.4 160 94 144 154 83 87

9297 5 139 79 134.6 138.1 76 77.7

2561 3.9 144 84 134.5 139.4 78.3 81.1

207 4 NR NR 132.4 140.4 82.3 86.5

857 1.1 182 100 151.9 174 83.6 94.5

1328 2 129 77 124 129.3 74.7 78.2

601 3.6 151 87 139 142 81 86

603 3.6 150 87 139 141 80 82

4912 3.9 145 82 143.5 145 81.3 82

8290 4.8 134 78 129.6 131.6 74.4 75.6

5269 3 136 83 127.9 132.1 78.6 81

35707 4.2 140.1 81.5 134.0 137.8 77.5 79.6

3544 3.9 149 87 133.1 145.2 79.1 84.1

600 3.6 151 87 139 142 80 83

11140 4.3 145 81 134.7 140.3 74.8 75.6

253 3.8 148 87 137.2 139.5 80.5 82.8

50944 4.2 141.9 81.9 134.1 138.9 77.1 79.3

included together with the entire group randomized to less intense treatment.with the matched group randomized to placebo has been included in the primaryopril and diuretic with the matched group randomized to placebo has been added in the

th the arm randomized to amlodipine only.randomized to placebo.

placebo.rsus verapamil have both been included in the primary analysis, whereas the comparison

patients being counted only once).as been included. The comparison delapril with placebo has not been included because

Volume 33 � Number 2 � February 2015

Outcome



Trials(n)

10106

126

1110

11117

157

1513


–4.2/–2.1–4.2/–2.1–3.5/–1.6–4.1/–2.1–3.5/–1.6–4.1/–2.0–4.1/–2.0

–5.1/–2.3–5.1/–2.3–4.1/–1.8–5.0/–2.3–4.1/–1.8–5.0/–2.3–5.0/–2.3

Treated

550/17058987/17058645/14942

1561/176591951/14952

633/163781274/15984

915/243971322/24397

865/205112274/254242651/205111055/254241990/24730

Controls

679/170361133/17036

822/149441834/176392377/14944

770/163591389/15958

1152/243811533/243811053/205152717/251113120/205151227/251112182/24710


(95% CI)

–9 (–13, –3)–9 (–14, –2)

–12 (–19, –4)–16 (–23, –9)

–29 (–42, –14)–5 (–12, +4)–7 (–16, +3)

–11 (–16, –5)–10 (–14, –4)–9 (–16, –2)

–20 (–27, –11)–23 (–36, –10)

–7 (–12, –1)–9 (–15, –2)

NNT5 years

(95% CI)

116 (76, 331)111 (69, 481)84 (52, 250)62 (44, 116)35 (24, 74)

182 (81, –285)135 (64, –360)

90 (62, 188)105 (70, 245)106 (63, 449)

49 (37, 93)43 (28, 101)

145 (86, 937)114 (69, 510)


(95% CI)

0.80 (0.69–0.93)0.87 (0.79–0.97)0.79 (0.66–0.93)0.85 (0.79–0.92)0.83 (0.75–0.92)0.89 (0.75–1.07)0.92 (0.83–1.03)

0.79 (0.69–0.90)0.86 (0.79–0.94)0.83 (0.71–0.96)0.83 (0.77–0.91)0.86 (0.78–0.94)0.87 (0.78–0.98)0.91 (0.85–0.98)


4.36.95.7

10.816.95.09.3

5.46.85.6

12.016.55.39.8

RR (95% CI)

P(Heterogen)

0.220.530.170.380.0590.0970.19

0.110.29

0.0820.0790.0080.160.33

Primary analysis

Secondary analysis

Angiotensin converting enzyme inhibitor better Control better

0.3 0.6 1.0 1.3

FIGURE 7 Relative and absolute risk reduction of various outcomes in the trials of blood pressure lowering by angiotensin-converting enzyme inhibitors. Primary andsecondary analyses include trials listed in Table 4. The column ‘Outcome risk’ reports the percentage incidence of each outcome in the control group calculated on a 5-year span. The column ‘Absolute risk reduction’ reports the number (and 95% CI) of events prevented every 1000 patients treated for 5 years with the observed RR. CHD,coronary heart disease; CI, confidence interval; CV, cardiovascular; HF, heart failure; n, number; NNT, number (and 95% CI) of patients needed to treat for 5 years toprevent one event; pts, patients; P (Heterogen), P for the heterogeneity test x2Q; RR, Mantel–Haenszel risk ratios.


diuretic, only slightly improved the risk reduction of mostoutcomes, but a 13% reduction in cardiovascular death anda 9% reduction in all-cause death achieved statistical sig-nificance (Fig. 7). Significant heterogeneity was found fortwo outcomes in the primary analysis and for three in thesecondary analysis.

Figure 7 also indicates the incidences of each outcome inthe control groups, with cardiovascular death risk averag-ing 5.0% in 5 years. Absolute risk reduction induced byACE-inhibitor therapy (primary analysis) amounted to 9strokes, 12 heart failure, nine CHD and 28 major cardio-vascular events (composite of strokes, CHD and heart fail-ure) every 1000 patients treated for 5 years (NNT 116, 84,111 and 35, respectively).


TABLE 5. BP lowering by angiotensin receptor blockers

Trial



Drugs compared

PanuActive Control

Trials for primary analysisRENAAL [60] 93 Yes Losartan Placebo 1

IDNT [45]a 100 Yes Irbesartan Placebo 1

IRMA-2 [61] 100 No Irbesartan Placebo

SCOPE [62] 100 Yes, low Candesartan Placebo 1

PROFESS [63] 74 Yes Telmisartan Placebo 2

TRANSCEND [64] 76.4 Yes Telmisartan Placebo 5

DIRECT-2 [65] 62 Yes Candesartan Placebo 1

I-PRESERVE [66] 88 Yes Irbesartan Placebo 4

GISSI-AF [67] 85.4 Yes Valsartan Placebo 1

NAVIGATOR [68] 77.5 Yes Valsartan Placebo 9

ACTIVE-I [69] 88 Yes Irbesartan Placebo 9

ROADMAP [70] 82 Yes Olmesartan Placebo 4

ORIENT [71] 93 Yes Olmesartan Placebo

Total primary 6

BP, blood pressure.aIn IDNT, comparison between the arm randomized to irbesartan and the arm randomized to p


The SBP/DBP difference between ACE inhibitors andcontrol therapy was much smaller in ACE-inhibitor basedthan in diuretic or beta-blocker initiated trials, probablybecause most of the trials were of nonintentional BP low-ering [1]. Risk ratio standardization to a 10/5 mmHg SBP/DBP difference was made, but should be seen as anuncertain approximation: stroke 0.52 (0.34–0.81); CHD0.65 (0.48–0.91); heart failure 0.47 (0.26–0.79); and com-posite of stroke, CHD and heart failure 0.54 (0.39–0.76).

Angiotensin receptor blockersTable 5 lists the 13 RCTs (65256 individuals) included in theanalysis of placebo-controlled trials, in which active treat-ment was based on an ARB (only primary analysis available).


tientmber

Follow-up(years)

BaselineBP

(mmHg)

AchievedSBP

(mmHg)

AchievedDBP

(mmHg)


513 3.4 152 82 143.5 146.2 76.7 77.7

148 2.6 159 87 140 144 77 80

590 2 153 90 142 144 83 83

937 3.7 166 90 145.2 148.5 79.9 81.6

0332 2.5 144 84 135.4 139.6 79.2 81.3

926 4.7 141 82 134.1 138.7 NR NR

905 4.7 133 77 NR NR NR NR

128 4.1 137 79 133.2 135.8 76.9 78.9

442 1 139 82 134 137 NR NR

306 6.5 140 83 133 136 78 82

016 4.1 138 82 131.5 134.3 78.1 79.6

447 3.2 136 80 126.7 128.7 74.3 76.2

566 3.2 139 76 132.5 137 73 74

5256 3.8 143.1 83.1 134.6 138.1 78.2 80.1

lacebo.


Thomopoulos et al.

The SBP/DBP difference between active and placebotreatments was rather small (about –3.7/–2.0 mmHg) andassociated with a significant relative reduction in stroke[–9% (–3 to –14%)], heart failure [–10% (–3 to –17%)] andmajor cardiovascular events [composite of stroke, CHD,heart failure –9% (–5 to –14%)], but not in CHD [–6%(þ4% to –14%)], cardiovascular death [þ3% (þ13 to – 6%)]and all-cause death [þ1% (þ6 to –3%)]. Only for cardio-vascular death, heterogeneity achieved statistical signifi-cance (Fig. 8).

In the control groups of ARB-based trials, incidence of alloutcomes, except CHD, was markedly higher than in trialsbased on all other classes of antihypertensive drugs, inci-dence of cardiovascular death averaging 7.9% in 5 years.Absolute risk reduction amounted to nine strokes, sevenheart failure and 20 major cardiovascular events (compositeof stroke, CHD and heart failure) every 1000 patients treatedfor 5 years (NNT 114, 144 and 50, respectively; Fig. 8).

Standardizations of risk ratios to 10/5 mmHg SBP/DBPdifferences were also calculated, but should be taken withcaution because of the large approximations: stroke 0.80(0.70–0.93); heart failure 0.75 (0.60–0.92); and compositeof stroke, CHD and heart failure 0.79 (0.69–0.88).

Renin–angiotensin system blockersTrials in the primary analyses of ACE inhibitors and ARBswere combined together (25 trials, 100 963 individuals).SBP/DBP differences between active and control treatmentof about –3.8/–2.0 mmHg were associated with significantreductions in stroke [–13% (–7 to –18%)], heart failure[–14% (–6 to 21%)], CHD [–10% (–3 to –15%)] and com-posite of stroke, CHD, and heart failure [–13% (–8 to�18%)], but not in cardiovascular mortality [–2% (þ8 to�10%)] and all-cause mortality [–2% (þ2 to –6%)]. Absoluterisk reductions amounted to nine strokes, 10 heart failure,six CHD and 27 major cardiovascular events every 1000patients treated for 5 years (NNT 109, 99, 180 and 37,respectively).

Centrally acting drugsOnly a secondary analysis of RCTs using centrally actingdrugs was possible, as six of the eight trials (but only 1786patients) listed in Table 6 administered a centrally acting

Outcome

Stroke CHD HF Stroke + CHD Stroke + CHD + HF CV Death All-cause Death

Trials (n)

11 11 8 10 8 10 13

Difference SBP/DBP (mmHg)

–3.7/–2.0 –3.5/–1.6 -3.8/–2.1 –3.7/–2.0 –3.8/–2.1 –3.7/–2.0 –3.7/–2.0

Treated

1734/31359 822/31026 1286/25928 2547/30637 3659/25928 1867/30637 3413/32699

Controls

1902/31402 868/30883 1429/26007 2775/30682 4100/26007 1833/30682 3377/32557

Absolute Risk Reduction 1000 pts/5 years

(95% CI)

–9 (–13, –3) –2 (–5, +1) –7 (–12, –2) –11 (–18, –4) –20 (–31, –11) +2 (–5, +10) +1 (–4, +9)

NNT 5 years

(95% CI)

114 (74, 334) 455 (198, –672)

144 (86, 475) 91 (57, 238) 50 (33, 90)

Events (n/patients) RR

(95% CI)

0.91 (0.86–0.97)0.94 (0.86–1.04)0.90 (0.83–0.97)0.92 (0.87–0.97)0.91 (0.86–0.95)1.03 (0.94–1.13)1.01 (0.97–1.06)

Outcome risk (% in 5 years)

10.1 3.7 7.0 14.1 22.5 7.9 14.4

RR (95% CI)

P (Heterogen)

0.68 0.48 0.36 0.88 0.22 0.078 0.92

Primary analysis

Angiotensin receptor blocker better Control better

0.3 0.6 1.0 1.3

IGURE 8 Relative and absolute risk reduction of various outcomes in trials of blood pressure lowering by angiotensin receptor blockers. Primary analyses include trialssted in Table 5. The column ‘Outcome risk’ reports the percentage incidence of each outcome in the control group calculated on a 5-year span. The column ‘Absolutesk reduction’ reports the number (and 95% CI) of events prevented every 1000 patients treated for 5 years with the observed RR. CHD, coronary heart disease; CI,onfidence interval; CV, cardiovascular; HF, heart failure; n, number; NNT, number (and 95% CI) of patients needed to treat for 5 years to prevent one event; pts, patients;(Heterogen), P for the heterogeneity test x2Q; RR, Mantel–Haenszel risk ratios.

FliricP


drug in combination with other drugs [29,30,33,34] or as achoice amongst various agents [31,32]. Only two trialsinitiated active treatment with a centrally acting drug only[72,73]. Most of the trials included in this analysis were earlyBP-lowering studies, patients had very high baseline and avery high incidence of cardiovascular outcomes in thecontrol group (average cardiovascular death 11.4% in5 years). SBP/DBP differences between active and placebotreatments were very large (–23/–14mmHg) and associ-ated with significant and marked reductions in all out-comes, except CHD and all-cause death. Stroke wasreduced by –47% (–14 to –68%), heart failure by –91%(–67 to –98%), major cardiovascular events (composite ofstroke, CHD and heart failure) by –46% (–14 to 66%) andcardiovascular death by 40% (–12 to –59%). A 13%reduction in CHD and a 17% reduction in all-cause deathwere not statistically significant (Fig. 9). Heterogeneity wassignificant for the composite of stroke, CHD and heartfailure, and for all-cause death.

As a result of the high risk of the patients and the markedBP reduction, absolute risk reductions were large: 73strokes, 35 heart failure, 108 major cardiovascular eventsand 41 cardiovascular deaths every 1000 patients treated5 years (NNT 14, 26, 9 and 24, respectively).

Influence of individual trials on pooled effectsizesIn none of the primary analyses excluding one trial at a timebrought the point estimate of the combined effect on anyoutcome outside the overall estimate with all availabletrials, indicating no trial had an excessive influence inany analysis.

DISCUSSION

Strengths and limitations of the evidenceconcerning the benefits of each class ofantihypertensive agentsA large number of trials comparing active BP-loweringtreatment based on a diuretic with placebo or no treatmentcould be included in the primary analysis (12 trials on littleless than 50 000 patients). Most of these trials (9 of 12) werecompleted in an early period of antihypertensive treatment


TABLE 6. BP lowering by centrally acting drugs

Trial



Drugs compared

Follow-up(years)

BaselineBP

(mmHg)

AchievedSBP

(mmHg)

AchievedDBP

(mmHg)

Active ControlPatientnumber SBP DBP Active Control Active Control

Trials for primary analysisWolff [72] 100 No Reserpine Placebo 87 1.4 177 109 157.1 190 95.5 115.3

VA-1 [29] 100 No Hydrochlorothiazide,reserpine andhydralazine

Placebo 143 1.5 186 121 142.6 182 91.9 118.7

VA-2 [30] 100 No Hydrochlorothiazideand reserpine andhydralazine

Placebo 380 3.8 164 104 134.9 169.3 86.4 105

Carter [31] 100 No Thiazide andmethyldopa ordebrisoquine

Untreated 99 4 >160 �110 NR NR 106 115

Barraclough [32] 100 No Methyldopa ordebrisoquine orbendrofluazide

Placebo 116 2 NR 110 NR NR 89.8 104.2

HSCSG [33] 100 No Methyclothiazideand deserpidine

Placebo 452 3 164 101 141 166 88 100

USPHS [34] 100 No Chlorthalidone andrauwolfia

Placebo 389 7 147 99 131.5 147.4 88.4 98.4

Sprackling [73] 100 No Methyldopa Untreated 120 5.5 201 108 180 196.5 99 103.3

Total primary 1786 4.0 165.2 104.3 141.2 167.3 90.2 104.3

BP, blood pressure.


(between 1978 and 1992), almost exclusively includedhypertensive patients with no background antihypertensivetherapypresent at randomizationandcontinuedover follow-up. Average baseline SBP and DBP were elevated (164/95mmHg), and on-treatment SBP/DBP differences betweenactive and placebo treatment were relatively large (about 12/5.5mmHg). Average cardiovascular risk in the control groupwas in the high category according to the current classifi-cations [15,74,75], with an incidence of cardiovascular deathof 8.8% in 10 years, and heterogeneity of the trials was low.All these characteristics gave high statistical power to themeta-analysis and favoured the demonstration that BP-low-ering therapy based on diuretics can significantly reduce alltypes of cardiovascular outcome, including cardiovasculardeath and all-cause death. Absolute risk reduction was alsoremarkable, with a NNT for 5 years to prevent one majorcardiovascular event of only 41 (34–54).


Outcome


Trials(n)

6757678


–23.4/–14.7–23.7/–14.6–23.8/–15.0–23.7/–14.6–23.2/–14.2–23.7/–14.6–23.2/–13.9

Treated

55/78024/8380/730

79/83897/79038/838

109/898

Controls

91/77028/82829/721

119/828157/78166/828

134/888

Events(n/patients)

(9

0.53 (0.30.87 (0.50.09 (0.00.66 (0.50.54 (0.30.60 (0.40.73 (0.4


17.93.85.020.525.111.418.0

Primary a

FIGURE 9 Relative and absolute risk reduction of various outcomes in trials of blood prTable 6. The column ‘Outcome risk’ reports the percentage incidence of each outcome ireduction’ reports the number (and 95% CI) of events prevented every 1000 patients treconfidence interval; CV, cardiovascular; HF, heart failure; n, number; NNT, number (andP (Heterogen), P for the heterogeneity test x2Q; RR, Mantel–Haenszel risk ratios.


The high statistical power of the group of BP-loweringRCTs investigating diuretics also allowed separate meta-analyses, showing a number of outcomes could be signifi-cantly reduced by low-dose diuretics (i.e. with doses cur-rently recommended) and drugs belonging to each of threemajor subclasses of diuretics (thiazides, chlorthalidone andindapamide).

Also, trials comparing centrally acting drugs withplacebo were mostly done in early times of antihyperten-sive treatment (1966–1981) in previously untreated hyper-tensive patients with high baseline SBP/DBP values (mean165/104mmHg), at very high cardiovascular risk (cardio-vascular death in the control group averaged 22.8% in10 years), and very large SBP/DBP differences betweenactive and control treatment (over 24/14 mmHg). However,these early trials were quite small and only a total of 1786patients could be included in this meta-analysis. In spite of


0.3 0.6 1.0 1.5


(95% CI)

–73 (–97, –24)–5 (–18, +20)–38 (–40, –30)–64 (–89, –30)

–108 (–149, –35)–41 (–57, –13)–47 (–89, +25)

NNT5 years

(95% CI)

14 (10, 41)202 (55, –50)

26 (25, 33)16 (11, 34)9 (7, 29)

24 (17, 75)21 (11, –39)

RR5% CI)

2–0.86)1–1.51)2–0.33)1–0.85)4–0.86)1–0.88)7–1.14)

RR (95% CI)

P(Heterogen)

0.190.800.920.580.0070.440.007

nalysis

Centrally acting drug better Control better

0.1

essure lowering by centrally acting drugs. Secondary analyses include trials listed inn the control group calculated on a 5-year span. The column ‘Absolute riskated for 5 years with the observed RR. CHD, coronary heart disease; CI,95% CI) of patients needed to treat for 5 years to prevent one event; pts, patients;


Thomopoulos et al.

the smallness of the population studied, a large and sig-nificant reduction of all major cardiovascular events, exceptCHD but including cardiovascular death (though not all-cause death), could be demonstrated. As a result of the highrisk of the population included, absolute risk reductionswere large, and NNT for 5 years to prevent one majorcardiovascular event was only 9. However, the results ofthe meta-analysis of centrally acting drugs must be takenwith caution, as in most of the trials centrally acting drugswere one of the possible choices of the investigator orwere administered in association with other agents sincerandomization.

Also, trials on beta-blockers versus placebo were con-ducted in early periods of antihypertensive therapy(between 1980 and 1998), but only 5 trials on less than19 000 patients could be included in the primary analysis.Although all individuals were hypertensive without base-line treatment continued over follow-up, SBP/DBP differ-ences between randomized treatments were rather large(about 10/7 mmHg) and for most outcomes trial hetero-geneity was low, the statistical power of the beta-blockermeta-analysis was lower than that in the diuretic meta-analysis. Despite this limitation, stroke, heart failure andmajor cardiovascular events (but not CHD) were signifi-cantly reduced by beta-blocker-based therapy, and whenthe STOP study [35], with two-thirds of the actively treatedpatients receiving a beta-blocker, was added in a sensitivityanalysis, also cardiovascular death was significantlyreduced. Absolute cardiovascular risk in the control groupwas in the high stratum (7.4% in 10 years) and absolute riskreduction by beta-blockers was substantial, with a NNT for5 years to prevent one major cardiovascular event of 64.

Several characteristics of the trials that have used morerecent antihypertensive drug classes are different fromthose of the trials that tested diuretics, centrally acting drugsand beta-blockers. In part because of the ethical reasonsresulting from the assumption that the benefits of BP low-ering had already been demonstrated by trials with earliercompounds, trials of more recent antihypertensive drugsoften enrolled hypertensive patients already on varioustypes of antihypertensive agents, which were continuedafter randomization on both arms of the trials, in such a waythat the drug to be tested and the respective placebo, ratherthan being the agents by which treatment was initiated,were the last ones to be added upon a background of otherantihypertensive drugs. Furthermore, some of the recenttrials (which we have defined of ‘nonintentional’ BP low-ering [1]) were aimed at showing BP-independent benefitsof new agents, and therefore enrolled also nonhypertensiveindividuals, although we tried to minimize this limitation byincluding in our meta-analysis only trials enrolling at least40% of hypertensive patients and excluding trials on acutemyocardial infarction and chronic heart failure. A majorconsequence of these differences in trial design was a muchsmaller BP difference between active and placebo arms.

These limitations affect to a lower degree the trials testingBP lowering by calcium antagonists (only data on dihy-dropyridines available). Of 10 trials (30 359 individuals)included in the primary analysis, six (20 222 individuals)exclusively enrolled hypertensive patients, three of whichused no background antihypertensive therapy (hence,


active therapy was initiated by a calcium antagonist) andthe fourth randomized to a calcium antagonist or placeboon the background of very low dose diuretic therapy.Nonetheless, SBP/DBP differences between active andplacebo therapies (�6/3 mmHg) were about half of thosein the trials of diuretics or beta-blockers. Despite thislimitation, BP lowering by calcium antagonists was foundto significantly reduce stroke, the composite of stroke andCHD and the composite of stroke, CHD and heart failure, aswell as both cardiovascular and all-cause mortality. A 17%reduction in CHD and a 19% reduction in heart failure,however, did not achieve statistical significance. A sensi-tivity analysis including only the four trials enrolling exclu-sively hypertensive patients and in most of which treatmentwas initiated by a calcium antagonist (i.e. with a designsimilar to trials testing diuretics, centrally acting drugs andbeta-blockers) showed that heterogeneity was reduced,and all outcomes (including CHD and heart failure) couldbe significantly reduced by BP lowering with calciumantagonists.

Level of total cardiovascular risk was comparable incalcium antagonist and diuretic or beta-blocker based trials(cardiovascular death in the control groups averaging 8.4,8.8 and 7.4% in 10 years, respectively), and absolute riskreduction induced by calcium antagonists, despite thelimited BP reduction versus placebo, was remarkable witha NNT for 5 years to prevent a major cardiovascular event of39 and of 107 to prevent a death.

The above-mentioned limitations of trials testing morerecent drug classes are particularly evident with trials inves-tigating ACE inhibitors. Nine of the twelve trials availablefor the primary analysis also included nonhypertensivepatients and in only two small studies treatment was reallyinitiated by an ACE inhibitor. In most cases the ACE inhibi-tor and the placebo were added to a background ofpreexisting antihypertensive therapy common to both armsof the trials. Consequently, the mean SBP/DBP differencesbetween ACE inhibitor and placebo treatments were verysmall (–4/–2mmHg), about one-third of those found withdiuretics and beta-blockers. Nonetheless, these small BPdifferences were accompanied by significant reductions inall types of cardiovascular events considered, but a 11% andan 8% reduction in cardiovascular mortality and, respec-tively, in all-cause mortality did not achieve statisticalsignificance. Whether this was because of the smallnessof the SBP/DBP differences obtained because the ACEinhibitor was, in most trials, the last antihypertensive agentadded to others cannot be decided. Both types of mortalitywere found to be significantly reduced (–13 and –9%,respectively) in the secondary meta-analysis, because ofthe addition of the large Action in Diabetes and VascularDisease (ADVANCE) trial [37] in which an ACE inhibitor wastested in fixed combination with a diuretic, and of thePerindopril Protection against Recurrent Stroke Study(PROGRESS) trial group [36] randomized to a combinationof ACE inhibitor and diuretic. Therefore, the achievedsignificance might be because of the concomitant use ofa diuretic. Total cardiovascular risk in the ACE-inhibitortrials was higher than in trials based on diuretics, beta-blockers or calcium antagonists (cardiovascular death incontrol groups being 10 rather than 8.8, 7.4 and 8.4% in



10 years), and absolute risk reductions of most cardio-vascular events were remarkable (with the exclusion offatal events), with a NNT for 5 years of 35 to prevent a majorcardiovascular event.

Trials that could be included in the meta-analysis of ARBstudies suffer the same limitations as those testing ACEinhibitors: 10 of 13 trials also enrolled nonhypertensivepatients, background antihypertensive therapy was contin-ued over randomized follow-up in 12 of the 13 trials and avery small SBP/DBP difference versus placebo wasobtained (–3.7/–2mmHg). Consequently, despite the verylarge number of individuals included in the meta-analysis(65 256 individuals), the small BP lowering produced byaddition of ARBs to previous therapy was associated onlywith significant reductions in strokes, heart failure and thecomposite of stroke and CHD, and of stroke, CHD and heartfailure, but not of CHD and of cardiovascular and all-cause mortalities.

The total cardiovascular risk of patients in the ARB meta-analysis was very high (cardiovascular death in the controlgroup averaged 15.8% in 10 years), and the absolute riskreduction of stroke, heart failure and major cardiovascularevent substantial, with a NNT for 5 years of 114, 144 and50, respectively.

It has to be mentioned that, even when ACE inhibitorand ARB based trials were pooled together (renin–angiotensin system blockers) with over 100 000 patientsincluded, both reductions in cardiovascular and all-causedeath did not achieve statistical significance.

A final difference between the trials testing the differentclasses of antihypertensive agents must be remarked.Mostly because of the trial design of adding the tested drugon a background of other antihypertensive drugs, largelyused with ACE inhibitors and ARBs, less frequently usedwith calcium antagonists and practically unknown withearlier antihypertensive agents, the baseline and theachieved levels of SBP/DBP in the various meta-analyseshere reported were highest with centrally acting drugs,followed by beta-blockers, diuretics, calcium antagonists,and were lowest with ACE inhibitors and ARBs. Although aprevious meta-analysis has reported that baseline BP valuesappear not to influence RR reduction [76], a more recent andlarger meta-analysis we have recently published [14] hasfound that at least the absolute benefit of BP lowering tendsto decrease when the same BP difference is obtained atlower BP levels, and at the lowest SBP values (across acutoff of 130 mmHg) only stroke is significantly reduced byBP-lowering therapy. This may be a further reason for thefailure so far to demonstrate the effectiveness of ACEinhibitors and ARBs to significantly reduce mortality.

Comparison with previous meta-analysesOurs is the largest and, at the same time, the most specificand rigorous meta-analysis of BP-lowering trials groupedaccording to the antihypertensive drug class that was com-pared to placebo, absence of treatment or usual treatment.It is the largest meta-analysis because it has considered allintentional and nonintentional BP-lowering trials, from thefirst one in 1966 to the end of 2013 [1], in which a BPdifference between the trial arms could be attributed to aspecific class of antihypertensive agents. It is the most

Copyright © 2015 Wolters KluwerJournal of Hypertension

specific meta-analysis because included trials enrolledeither exclusively hypertensive patients or at least 40% ofpatients with hypertension, and because all trials conductedafter an acute myocardial infarction, left ventricular dys-function, heart failure and acute stroke were excluded. It isthe most rigorous meta-analysis because the temptation tocompare the various meta-analyses testing different classesin order to estimate their different effectiveness on variousoutcomes has been resisted; indeed, as remarked above,the characteristics of the patients included in trials usingdifferent drug classes were quite different, the design of thetrials different (the drug confronted with placebo being theone by which treatment was initiated in trials on diureticsand beta-blockers, and in most of those on calciumantagonists, whereas it was the final one added on a back-ground of other antihypertensive agents in most trials onblockers of the renin–angiotensin system), the BP differ-ence achieved between the active and control arms varyingbetween –23 and –3.5 mmHg SBP, –12.6 and –1.6 mmHgDBP according to the drug class considered. Finally, theonly correct design to evaluate the differences in theeffectiveness of different therapeutic regimens is head-to-head randomized comparison of different regimens.

As a result of the characteristics mentioned above, ourmeta-analyses differ from the four major groups of meta-analyses published so far. The major difference from themeta-analysis by Psaty et al. [2] is not so much in the numberof trials we could include (18 of the trials we have includedwere published after closure of Psaty et al.’s survey at theend of 2002), but in the fact that Psaty et al. constructed aso-called network meta-analysis by putting together thetrials testing different regimens, provided that they had onetreatment in common. Although we understand networkmeta-analyses are based upon sophisticated statisticalassumptions, this large body of assumptions and the con-sequent adjustments make – in our opinion – indirectcomparisons much less valuable clinically than the directcomparisons we have chosen to make. The Blood PressureLowering Treatment Trialists’ Collaboration (BPLTTC), hav-ing chosen to make prospective meta-analyses, haveexcluded all trials published before the establishment oftheir protocol in 1997 [77], and therefore do not report anymeta-analysis of placebo-controlled trials using diuretics orbeta-blockers [3,4]. The last BPLTTC meta-analysis report-ing comparisons of calcium antagonists and ACE inhibitorswith placebo is a recent one on the influence of chronickidney disease: it includes only three trials comparing acalcium antagonist with placebo (one not included in ourmeta-analysis because in predominantly nonhypertensivepatients), and 10 trials comparing an ACE inhibitor withplacebo (5 of which not included in our primary meta-analysis because of predominantly nonhypertensivepatients with CHD or atherosclerosis or because testingan ACE inhibitor in fixed combination with a diuretic [78]).Law et al.’s meta-analysis [8], apart from not including trialspublished after 2007, is quite indiscriminate as far as itconsiders a large number of trials using antihypertensiveagents in conditions different from hypertension: in thecalcium antagonist versus placebo meta-analysis, sevenhypertension trials are diluted by data from 15 other trialsmostly on acute myocardial infarction, heart failure and

Health, Inc. All rights reserved.www.jhypertension.com 207

Thomopoulos et al.

other conditions different from hypertension; in the ACEinhibitor versus placebo trials, data from seven trials alsoincluded in our meta-analysis are confounded by data fromother 15 trials on acute myocardial infarction, heart failure,etc.; the ARB versus placebo meta-analysis is limited to fourtrials in patients with CHD and heart failure rather thanhypertension. The design of the 2009 Cochrane Collabor-ation meta-analyses [10] is rather similar to ours as far asdiuretics are concerned, though excluding an importantlarge trial such as the Hypertension Detection Follow-upProgram (HDFP) [18,19] and including trials that we haveconsidered for a secondary meta-analysis only, because adiuretic was given together or as an alternative to centrallyacting drugs. Regarding more recent antihypertensive drugclasses, the 2009 Cochrane collaboration [10] could onlyconsider one trial comparing a calcium antagonist withplacebo, three comparing ACE inhibitors with placeboand none comparing ARBs with placebo.

A few authors have focused on the meta-analyses ofplacebo-controlled trials using a given class of drugs. Intheir seminal meta-analysis on beta-blockers, Lindholmet al. [5] have included the five major trials we have alsoconsidered, inclusive of STOP [20,25,35,38,39], but exclu-sive of UK Prospective Diabetes Study (UKPDS) [40,41], buthave also considered Dutch Transient Ischemic Attack(Dutch TIA) [79] (which we have excluded because ofuncertainty about the proportion of hypertensives) andIPPPSH [80] (which we have considered a trial comparingactive regimens). The 2012 Cochrane Collaboration meta-analyses by Wiysong et al. [13] has considered only fourbeta-blocker initiated trials, three of which [20,25,38] wereamongst the six we have included, and the InternationalProspective Primary Prevention Study in Hypertension(IPPPSH) trial [80]. Costanzo et al. [9] in 2009 meta-analysesof trials on calcium antagonists analyzed together theplacebo-controlled and active-controlled trials; when sep-arate analyses were done, only four placebo-controlledtrials were included [43,48,49,51] (rather than 10 consideredin our meta-analysis). Van Vark et al. [12] focused on trialsusing ACE inhibitors or ARBs, but they also mixed placebo-controlled and active-controlled trials, an unsuitable designfor estimating the BP-lowering benefits obtained by the useof a specific class of drugs. Furthermore, inclusion in theACE-inhibitor meta-analysis of the Hypertension in the VeryElderly Trial (HYVET) [28], initiated by a diuretic with thesubsequent optional addition of an ACE inhibitor, and ofthe ADVANCE [37] trial, in which patients in the activegroup received a fixed combination of an ACE inhibitor anda diuretic, is disputable. Finally, Bangalore et al. [11] in theirmeta-analyses of the effects of ARB treatment, thoughcorrectly analyzing separately placebo-controlled andactive-controlled trials, have analyzed together data from10 (of 13) placebo-controlled trials on predominantlyhypertensive patients, with data from seven trials on non-hypertensive conditions, mostly heart failure.

For the reasons discussed above, our quantitative esti-mations of the drug class effects on different outcomes canonly be compared with those few meta-analyses that havefollowed the selection criteria similar to or not excessivelydissimilar from ours. As far as diuretics are concerned, theevidence we have found of the beneficial effects of BP


lowering by this class of drugs on all types of fatal andnonfatal outcomes, and the quantitative estimation of thesebenefits are very similar to what was reported by the 2009Cochrane Collaboration [10], both meta-analyses showingall types of outcomes are significantly reduced by diureticswith extremely similar risk ratios. Although we have used adefinition of low diuretic dose based on the maximum doseallowed by the protocol rather than the initial dose, also ourmeta-analyses show low-dose diuretics, as currently recom-mended, are associated with significant reduction of alloutcomes including cardiovascular and all-cause death. Asthe Cochrane collaboration [10], we have found onlystrokes and the composite of major cardiovascular events,but not mortality, are found significantly reduced by high-dose diuretics, but suggest this might be because of thequite lower cardiovascular risk in the trials using high-dosediuretics. In addition, we have separately analyzed diuretic-based trials according to the subclass of diuretic being used(thiazide, chlorthalidone or indapamide) and found a num-ber of clinically important outcomes can be reduced by anyof these diuretic subclasses, even at low dose, thus support-ing the recent European recommendations that any sub-class of diuretic can be prescribed to hypertensive patients[75]. As to beta-blockers, our estimations agree with themeta-analyses of Lindholm et al. [5] and the two CochraneCollaborations [10,13] so far as stroke and major cardiovas-cular events, but not CHD and all-cause mortality, cansignificantly be reduced, compared with placebo, by BPlowering with a beta-blocker. Our estimation of the beta-blocker effect on cardiovascular mortality (not calculatedby the three previous meta-analyses [5,10,13]) reveals that,when the STOP trial [35] is included, cardiovascularmortality is significantly reduced by beta-blockers. Noreasonable comparison can be done with the previousmeta-analyses regarding BP lowering by calciumantagonists, ACE inhibitors or ARBs because of the toolimited number of trials the previous meta-analyses haveconsidered or the different criteria they have followed.

CONCLUSIONOur meta-analyses show that BP lowering induced by allspecific classes of antihypertensive drugs is accompaniedby a significant reduction of the relative and absolute risk ofstroke and major cardiovascular events, and support theconcept that reduction of these events is because of BPlowering per se rather than to specific properties of thevarious classes of agents [75]. However, evidence of riskreduction of other cardiovascular events and, particularly,of mortality differs from class to class. Only for BP loweringby diuretics, there is evidence of significant reduction of allfatal and nonfatal outcomes. For beta-blockers, reductionsin CHD and all-cause mortality do not achieve statisticalsignificance, although cardiovascular mortality appears tobe significantly reduced when the STOP trial is included.For calcium antagonists, no statistically significantreduction in CHD and heart failure has been found, butstroke and both cardiovascular and all-cause mortality aresignificantly reduced. However, when more homogeneoustrials are considered (only hypertensive patients and no orminimal baseline antihypertensive treatment), also CHD



and heart failure are found to be significantly reduced bycalcium antagonists. Finally, no evidence of significantreduction in mortality (both cardiovascular and all-cause)has been achieved by BP lowering using ACE inhibitors andARBs. Likewise, evidence of significant reductions in therisk of CHD and all-cause mortality (but not cardiovascularmortality) is lacking for the centrally acting drugs.

The above-mentioned differences in the evidence avail-able for every drug class cannot be taken to mean that BPlowering by different classes of agents has partly differenteffectiveness on the risk of cardiovascular outcomes.Indeed, there are good reasons that may be suggested toexplain the differences we have described. Trials usingcentrally acting drugs were all very small and the power ofthe meta-analysis low. The same is the case of beta-blocker-based trials, only 6 trials on 28 000 patients beingavailable (to be compared to 12 trials on almost 50 000patients for diuretics), and the upper confidence limits onlyslightly above unity suggest a larger number of patientsmay have led to reach statistical significance also forreductions in CHD and total mortality. BP differencesbetween active and placebo treatments were quite smallin trials using calcium antagonists and, particularly, ACEinhibitors and ARBs. It is not unreasonable (though unpro-ven) to suppose that a larger BP reduction with respect toplacebo may have significantly reduced the mortalityoutcomes also.

Possible differences in the effectiveness of differentclasses of antihypertensive agents at least on some out-comes can only be estimated by head-to-head comparisonof two different classes of agents. However, the meta-analyses of class-specific placebo-controlled trials we havepresented here must be considered a necessary back-ground for the interpretation of the results of trials compar-ing head-to-head different active therapeutic regimens andof their meta-analyses (Thomopoulos, Parati, Zanchetti, inpreparation). Often, similar incidences of a given outcomein active regimens comparative trials are interpreted assimilar ‘benefits’ of the two active regimens. The data wehave reported here are a warning that similar outcomeincidences can be defined as ‘benefits’ only when at leastone of the agents tested has been shown to significantlyreduce the risk of that outcome in placebo-controlled trials.This we have shown here not to be true, at least at themoment, for all classes of drugs and for all outcomes.

ACKNOWLEDGEMENTSThe valuable help of Mrs Donatella Mihalich for literaturesearching and manuscript preparation, and that of Mrs.Paulina Wijnmaalen for the preparation of illustrationsare gratefully acknowledged. A.Z. is responsible for thedesign of the study and preparation of the first draft of themanuscript, A.Z. and C.T. have done the systematic reviewof the literature and extracted data, C.T. has conducted themeta-analyses, but all the three authors (C.T., G.P. and A.Z.)have substantially contributed to the interpretation of data,critical revision of the manuscript for important intellectualcontent, and given final approval of the version to bepublished. A.Z. and C.T. take responsibility of the integrityof the analyses.


Financial support and sponsorshipThe overview and meta-analyses were designed and con-ducted with funds made available to Istituto AuxologicoItaliano by current research grants of the Ministry of Healthof Italy, and in the context of contract EC 278249 (EU-MASCARA). C.T. was a visiting investigator at the IstitutoAuxologico Italiano, Milan, Italy, under a fellowshipgranted by the Hellenic Society of Cardiology.

Conflicts of interestThe authors declare no conflicts of interest regarding theoverview and meta-analyses, but C.T. declares consultancyfees from Astra Zeneca and lecture honoraria from Sanofi;G.P. declares lecture honoraria from Bayer, Daiichi Sankyo,Guidotti and Boehringer Ingelheim; and A.Z. declareslecture honoraria from Menarini International, RecordatiSpA and CVRx.

REFERENCES1. Thomopoulos C, Parati G, Zanchetti A. Effects of blood pressure

lowering on outcome incidence in hypertension. 1. Overview, meta-analyses and meta-regression analyses of randomized trials. J Hyper-tens 2014; 32:2285–2295.

2. Blood Pressure Lowering Treatment Trialists’ Collaboration. Effects ofACE inhibitors, calcium antagonists, and other blood-pressure-lowering drugs: results of prospectively designed overviews ofrandomised trials. Lancet 2000; 356:1955–1964.

3. Blood Pressure Lowering Treatment Trialists’ Collaboration. Effects ofdifferent blood-pressure-lowering regimens on major cardiovascularevents: results of prospectively-designed overviews of randomisedtrials. Lancet 2003; 362:1527–1535.

4. Blood Pressure Lowering Treatment Trialists’ Collaboration. Effects ofdifferent blood pressure-lowering regimens on major cardiovascularevents in individuals with and without diabetes mellitus: results ofprospectively designed overviews of randomized trials. Arch InternMed 2005; 165:1410–1419.

5. Lindholm LH, Carlberg B, Samuelsson O. Should beta blockers remainfirst choice in the treatment of primary hypertension? A meta-analysis.Lancet 2005; 366:1545–1553.

6. Reboldi G, Gentile G, Angeli F, Ambrosio G, Mancia G, Verdecchia P.Effects of intensive blood pressure reduction on myocardial infarctionand stroke in diabetes: a meta-analysis in patients. J Hypertens 2011;29:1253–1269.

7. Verdecchia P, Reboldi G, Angeli F, Gattobigio R, Bentivoglio M, Thijs L,et al. Angiotensin-converting enzyme inhibitors and calcium channelblockers for coronary heart disease and stroke prevention. Hyperten-sion 2005; 46:386–392.

8. Law MR, Morris JK, Wald NJ. Use of blood pressure lowering drugs inthe prevention of cardiovascular disease: meta-analysis of 147 random-ised trials in the context of expectations from prospective epidemio-logical studies. BMJ 2009; 338:b1665; doi: 10.1136/bmj.b1665.

9. Costanzo P, Perrone-Filardi P, Petretta M, Marciano C, Vassallo E,Gargiulo P, et al. Calcium channel blockers and cardiovascular out-comes: a meta-analysis of 175634 patients. J Hypertens 2009; 27:1136–1151.

10. Wright JM, Musini VM. First-line drugs for hypertension. CochraneDatabase Syst Rev 2009; doi: 10.1002/14651858.CD001841.pub2.

11. Bangalore S, Kumar S, Wetteslev J, Messerli FH. Angiotensin receptorblockers and risk of myocardial infarction: meta-analyses and trialsequential analyses of 147020 patients from randomized trials. BMJ2011; 342:d2234; doi: 10.1136/bmj.d2234.

12. Van Vark LC, Bertrand M, Akkerhuis KM, Brugts JJ, Fox K, Mourad J-J,et al. Angiotensin-converting enzyme inhibitors reduce mortality inhypertension: a meta-analysis of randomized clinical trials of renin–angiotensin–aldosterone system inhibitors involving 158998 patients.Eur Heart J 2012; 33:2088–2097.

13. Wiysonge CS, Bradley HA, Volmink J, Mayosi BM, Mbewu A, Opie LH.Beta-blockers of hypertension. Cochrane Lib 2012; doi: 10.1002/14651858.CD002003.pub4.


Thomopoulos et al.

14. Thomopoulos C, Parati G, Zanchetti A. Effects of blood pressurelowering on outcome incidence in hypertension. 2. Effects at differentbaseline and achieved blood pressure levels. Overview and meta-analyses of randomized trials. J Hypertens 2014; 32:2296–2304.

15. Thomopoulos C, Parati G, Zanchetti A. Effects of blood pressurelowering on outcome incidence in hypertension. 3. Effects in patientsat different levels of cardiovascular risk. Overview and meta-analysesof randomized trials. J Hypertens 2014; 32:2305–2314.

16. Perry HM Jr, Goldman AI, Lavin MA, Schnaper HW, Fitz AE, FrohlichED, et al. Evaluation of drug treatment in mild hypertension: VA-NHLBIfeasibility trial. Plan and preliminary results of a two-year feasibility trialfor a multicenter intervention study to evaluate the benefits versus thedisadvantages of treating mild hypertension. Prepared for the VeteransAdministration-National Heart, Lung, and Blood Institute Study Groupfor Evaluating Treatment in Mild Hypertension. Ann N Y Acad Sci 1978;304:267–292.

17. Management Committee. The Australian therapeutic trial in mild hy-pertension. Lancet 1980; 315:1261–1267.

18. Hypertension Detection Follow-up Program Cooperative Group.Five-year findings of the hypertension detection and follow-up pro-gram. I. Reduction in mortality of persons with high blood pressure,including mild hypertension. JAMA 1979; 242:2562–2571.

19. Davis BR, Ford CE. The hypertension detection follow-up program. In:Black HR, editor. Clinical trials in hypertension. New York: MarcelDekker, Inc; 2001. pp. 27–60.

20. Medical Research Council Working Party. MRC trial on treatment ofmild hypertension: principal results. BMJ 1985; 291:97–104.

21. Helgeland A. Treatment of mild hypertension: a five year controlleddrug trial. The Oslo study. Am J Med 1980; 69:725–732.

22. Amery A, Brixko P, Clement D, De Schaepdryver A, Fagard R, Forte J,et al. Mortality and morbidity results from the European Working Partyon High Blood Pressure in the Elderly trial. Lancet 1985; 325:1349–1354.

23. Perry HM Jr, Smith WM, McDonald RH, Black D, Cutler JA, Furberg CD,et al. Morbidity and mortality in the Systolic Hypertension in the ElderlyProgram (SHEP) pilot study. Stroke 1989; 20:4–13.

24. SHEP Co-operative Research Group. Prevention of stroke by antihy-pertensive drug treatment in older persons with isolated systolic,hypertension. Final results of the Systolic Hypertension in the ElderlyProgram (SHEP). JAMA 1991; 265:3255–3264.

25. MRC Working Party. Medical Research Council trial of treatment ofhypertension in older adults: principal results. BMJ 1992; 304:405–412.

26. PATS Collaborating Group. Poststroke antihypertensive treatmentstudy. A preliminary result. Chin Med J (Engl) 1995; 108:710–717.

27. Bulpitt CJ, Beckett NS, Cooke J, Dumitrascu D, Gil-Extremera B,Nachev C, et al. Hypertension in the Very Elderly Trial Working Group.Results of the pilot study for the Hypertension in the Very Elderly Trial.J Hypertens 2003; 21:2409–2417.

28. Beckett NS, Peters R, Fletcher AE, Staessen JA, Liu L, Dumitrascu D,et al. Treatment of hypertension in patients 80 years of age or older.N Engl J Med 2008; 358:1887–1898.

29. Veterans Administration Cooperative Study Group on Antihyperten-sive Agents. Effects of treatment on morbidity in hypertension. Resultsin patients with diastolic blood pressures averaging 115 through129mmHg. JAMA 1967; 202:1028–1034.

30. Veterans Administration Cooperative Study Group on Antihyperten-sive Agents. Effects of treatment on morbidity in hypertension. Resultsin patients with diastolic blood pressures averaging 90 through114mmHg. JAMA 1970; 213:1143–1152.

31. Carter BA. Hypotensive therapy in stroke survivors. Lancet 1970;295:485–489.

32. Barraclough M, Bainton D, Cochrane AL. Control of moderately raisedblood pressure. Report of a co-operative randomized controlled trial.BMJ 1973; 3:434–436.

33. Hypertension–Stroke Cooperative Study Group. Effect of antihyper-tensive treatment on stroke recurrence. JAMA 1974; 229:409–418.

34. Smith WM. Treatment of mild hypertension: results of a ten-yearintervention trial. Circ Res 1977; 40:I98–I105.

35. Dahlof B, Lindholm LH, Hansson L, Schersten B, Ekbom T, Wester PO.Morbidity and mortality in the Swedish Trial in Old Patients withHypertension (STOP-Hypertension). Lancet 1991; 338:1281–1285.

36. Collaborative PROGRESS Group. Randomised trial of a perindoprilbased blood-pressure-lowering regimen among 6105 individuals withprevious stroke or transient ischaemic attack. Lancet 2001; 358:1033–1041.


37. Collaborative ADVANCE Group. Effects of a fixed combination ofperindopril and indapamide on macrovascular and microvascularoutcomes in patients with type 2 diabetes mellitus (the ADVANCEtrial): a randomised controlled trial. Lancet 2007; 37:829–840.

38. Coope J, Warrender TS. Randomised trial of treatment of hypertensionin elderly patients in primary care. BMJ 1986; 293:1145–1151.

39. Smith WM. Treatment of mild hypertension: results of a ten-yearintervention trial. Circ Res 1977; 40:I98–105.

40. UK Prospective Diabetes Study Group. Tight blood pressure controland risk of macrovascular and microvascular complications in type 2diabetes: UKPDS 38. BMJ 1998; 317:703–713.

41. UK Prospective Diabetes Study Group. Efficacy of atenolol and capto-pril in reducing risk of macrovascular and microvascular complicationsin type 2 diabetes: UKPDS 39. BMJ 1998; 317:713–720.

42. Sun M, Zhou H, Jia Z. Prevention and treatment of stroke afterhypertension for ten years in Hunan Province. Zhonghua Nei Ke ZaZhi 1997; 36:312–314.

43. Staessen JA, Fagard R, Thijs L, Celis H, Arabidze GG, Birkenhager WH,et al. Randomised double-blind comparison of placebo and activetreatment for older patients with isolated systolic hypertension. TheSystolic Hypertension in Europe (Syst-Eur) Trial Investigators. Lancet1997; 350:757–764.

44. Liu L, Wang JG, Gong L, Liu G, Staessen JA. Comparison of activetreatment and placebo in older Chinese patients with isolated systolichypertension. Systolic Hypertension in China (Syst-China) Collabora-tive Group. J Hypertens 1998; 16:1823–1829.

45. Lewis EJ, Hunsicker LG, Clarke WR, Berl T, Pohl MA, Lewis JB, et al.Renoprotective effect of the angiotensin-receptor antagonist irbesartanin patients with nephropathy due to type 2 diabetes. N Engl J Med 2001;345:851–860.

46. Fogari R, Preti P, Zoppi A, Rinaldi A, Corradi L, Pasotti C, et al. Effects ofamlodipine fosinopril combination on microalbuminuria in hyper-tensive type 2 diabetic patients. Am J Hypertens 2002; 15:1042–1049.

47. Dens JA, Desmet WJ, Coussement P, De Scheerder IK, Kostopoulos K,Kerdsinchai P, et al. Long term effects of nisoldipine on the progressionof coronary atherosclerosis and the occurrence of clinical events: theNICOLE study. Heart 2003; 89:887–892.

48. Nissen SE, Tuzcu EM, Libby P, Thompson PD, Ghali M, Garza D, et al.Effect of antihypertensive agents on cardiovascular events in patientswith coronary disease and normal blood pressure: the CAMELOTstudy: a randomized controlled trial. JAMA 2004; 292:2217–2225.

49. Poole-Wilson PA, Lubsen J, Kirwan BA, van Dalen FJ, Wagener G,Danchin N, et al. Effect of long-acting nifedipine on mortality andcardiovascular morbidity in patients with stable angina requiring treat-ment (ACTION trial): randomised controlled trial. Lancet 2004;364:849–857.

50. Ruggenenti P, Perna A, Loriga G, Ganeva M, Ene-Iordache B, TurturroM, et al. Blood-pressure control for renoprotection in patients withnondiabetic chronic renal disease (REIN-2): multicentre, randomizedcontrolled trial. Lancet 2005; 365:939–946.

51. Liu L, Zhang Y, Liu G, Li W, Zhang X, Zanchetti A. The FelodipineEvent Reduction (FEVER) Study: a randomized long-term placebocontrolled trial in Chinese hypertensive patients. J Hypertens 2005;23:2157–2172.

52. Ruggenenti P, Fassi A, Ilieva AP, Bruno S, Iliev IP, Brusegan V, et al.Bergamo Nephrologic Diabetes Complications Trial (BENEDICT)Investigators. Preventing microalbuminuria in type 2 diabetes. N EnglJ Med 2004; 351:1941–1951.

53. Ruggenenti P, Lauria G, Iliev IP, Fassi A, Ilieva AP, Rota S, et al., for theDEMAND Study Investigators. Effects of manidipine and delapril inhypertensive patients with type 2 diabetes mellitus: The Delapril andManidipine for Nephroprotection in Diabetes (DEMAND) randomizedclinical trial. Hypertension 2011; 58:776–783.

54. Lewis EJ, Hunsicker LG, Bain RP, Rohde RD. The effect of angiotensinconverting-enzyme inhibition on diabetic nephropathy. The Collabo-rative Study Group. N Engl J Med 1993; 329:1456–1462.

55. Maschio G, Alberti D, Janin G, Locatelli F, Mann JFE, Motolese M, et al.Effect of the angiotensin-converting-enzyme inhibitor benazepril onthe progression of chronic renal insufficiency. The Angiotensin-Con-verting-Enzyme Inhibition in Progressive Renal Insufficiency StudyGroup. N Engl J Med 1996; 334:939–945.

56. The Heart Outcomes Prevention Evaluation Study Investigators. Effectsof an angiotensin-converting-enzyme inhibitor, ramipril, on cardiovas-cular events in high-risk patients. N Engl J Med 2000; 342:145–153.



57. Marre M, Lievre M, Chatellier G, Mann JF, Passa P, Menard J, DIA-BHYCAR Study Investigators. Effects of low dose ramipril on cardio-vascular and renal outcomes in patients with type 2 diabetes and raisedexcretion of urinary albumin: randomised, double blind, placebocontrolled trial (the DIABHYCAR study). BMJ 2004; 328:495.

58. The PEACE Trial Investigators. Angiotensin-converting-enzyme inhi-bition in stable coronary artery disease. N Engl J Med 2004; 351:2058–2068.

59. Trial DREAM Investigators. Effects of ramipril and rosiglitazone oncardiovascular and renal outcomes in people with impaired glucosetolerance or impaired fasting glucose: results of the Diabetes REductionAssessment with ramipril and rosiglitazone Medication (DREAM) trial.Diabetes Care 2008; 31:1007–1014.

60. Brenner BM, Cooper ME, de Zeeuw D, Keane WF, Mitch WE, ParvingHH, et al. Effects of losartan on renal and cardiovascular outcomes inpatients with type 2 diabetes and nephropathy. N Engl J Med 2001;345:861–869.

61. Parving HH, Lehnert H, Brochner-Mortensen J, Gomis R, Andersen S,Arner P. Irbesartan in Patients with Type 2 Diabetes and Microalbu-minuria Study Group. The effect of irbesartan on the development ofdiabetic nephropathy in patients with type 2 diabetes. N Engl J Med2001; 345:870–878.

62. Lithell H, Hansson L, Skoog I, Elmfeldt D, Hofman A, Olofsson B, et al.The Study on Cognition and Prognosis in the Elderly (SCOPE): prin-cipal results of a randomized double-blind intervention trial. J Hyper-tens 2003; 21:875–886.

63. Yusuf S, Diener HC, Sacco RL, Cotton D, Ounpuu S, Lawton WA, et al. ,Telmisartan to prevent recurrent stroke and cardiovascular events.N Engl J Med 2008; 359:1225–1237.

64. Telmisartan Randomized Assessment Study in ACE intolerant subjectswith cardiovascular disease (TRANSCEND) Investigators. Effects of theangiotensin-receptor blocker telmisartan on cardiovascular events inhigh-risk patients intolerant to angiotensin-converting enzyme inhibi-tors: a randomised controlled trial. Lancet 2008; 372:1174–1183.

65. Sjølie AK, Klein R, Porta M, Orchard T, Fuller J, Parving HH, et al. Effectof candesartan on progression and regression of retinopathy in type 2diabetes (DIRECT-Protect 2): a randomised placebo-controlled trial.Lancet 2008; 372:1385–1393.

66. Massie BM, Carson PE, McMurray JJ, Komajda M, McKelvie R, Zile MR,et al. Irbesartan in patients with heart failure and preserved ejectionfraction. N Engl J Med 2008; 359:2456–2467.

67. The GISSI-AF Investigators. Valsartan for prevention of recurrent atrialfibrillation. N Engl J Med 2009; 360:1606–1617.

68. The NAVIGATOR Study Group. Effect of valsartan on the incidence ofdiabetes and cardiovascular events. N Eng J Med 2010; 362:1477–1490.

69. The Active I Investigators. Irbesartan in patients with atrial fibrillation.N Engl J Med 2011; 364:928–938.


70. Haller H, Ito S, Izzo JL Jr, Januszewicz A, Katayama S, Menne J, et al.,ROADMAP Trial Investigators. Olmesartan for the delay or preventionof microalbuminuria in type 2 diabetes. N Engl J Med 2011; 364:907–917.

71. Imai E, Chan JC, Ito S, Yamasaki T, Kobayashi F, Haneda M, Makino H,for the ORIENT Study Investigators. Effects of olmesartan on renal andcardiovascular outcomes in type 2 diabetes with overt nephropathy: amulticentre, randomised, placebo-controlled study. Diabetologia 2011;54:2978–2986.

72. Wolff FW, Lindeman RD. Effects of treatment in hypertension. Resultsof a controlled study. J Chronic Dis 1966; 19:227–240.

73. Sprackling ME, Mitchell JR, Short AH, Watt G. Blood pressure reductionin elderly: a randomised controlled trial of methyldopa. BMJ 1981;283:1151–1153.

74. Perk J, De Backer G, Gohlke H, Graham I, Reiner Z, Verschuren M,et al. European Guidelines on cardiovascular disease prevention inclinical practice (version 2012): The Fifth Joint Task Force of theEuropean Society of Cardiology and Other Societies on CardiovascularDisease Prevention in Clinical Practice (constituted by representativesof nine societies and by invited experts) Developed with the specialcontribution of the European Association for Cardiovascular Preven-tion & Rehabilitation (EACPR). Eur Heart J 2012; 33:1635–1701.

75. Mancia G, Fagard R, Narkiewicz K, Redon J, Zanchetti A, Bohm M, et al.The Task Force for the management of arterial hypertension of theEuropean Society of Hypertension (ESH) and of the European Societyof Cardiology (ESC). 2013ESH/ESC Guidelines for the management ofarterial hypertension. J Hypertens 2013; 31:1281–1357.

76. Czernichow S, Zanchetti A, Turnbull F, Barzi F, Ninomiya T, Kengne A-P, et al. The effects of blood pressure reduction and of different bloodpressure-lowering regimens on major cardiovascular events accordingto baseline blood pressure: meta-analysis of randomized trials. J Hyper-tens 2011; 29:4–16.

77. Protocol for prospective collaborative overviews of major randomizedtrials of blood-pressure lowering treatments. World Health Organiz-ation–International Society of Hypertension Blood Pressure LoweringTreatment Trialists’ Collaboration. J Hypertens 1998; 16:127–137.

78. Blood Pressure Lowering Treatment Trialists’ Collaboration. Bloodpressure lowering and major cardiovascular events in people withand without chronic kidney disease: meta-analysis of randomizedcontrolled trials. BMJ 2013; 347:f5680; doi:10. 1136/bmj.f5680.

79. The Dutch TIA Trial Study Group. Predictors of major vascular eventsin patients with a transient ischemic attack or nondisabling stroke.Stroke 1993; 24:527–531.

80. The IPPPSH Collaborative Groups. Cardiovascular risk and risk factorsin a randomized trial of treatment based on the beta-blocker oxpre-nolol: the International Prospective Primary Prevention Study in Hy-pertension (IPPPSH). J Hypertens 1985; 3:379–392.


effectsofbloodpressureloweringonoutcome …...drugs of different pharmacological classes. there is...

Documents